Embed Size (px)
Citation preview
D-A137 592 MIXED MODE FOR GROUP 4 FACSIMILE SYSTEMS(U) DELTA 1/2INFORMATION SYSTEMS INC JENKINTOWN PA A DEUTERMANN07 NOV 83 NCS-TIB-83-7 DCA1iO-82-C-0047
UNCLASSIFIED F/G 14/5 NLE~llEEEEEEElll
lllslllllmllllEllhEEEEEElhlEEIIIEIIIIEEIIEElhEEEEEEElllI
.
12 IIIL.4 1116
MICROCOPY RESOLUTION TEST CHARTNATICC4A BUREAU OF STANDARDS-1963-A
Af
AM ON - - 9 W
%t.
%.
NCS TIB 83-7
NATIONAL COMMUNICATIONS SYSTEM
*'
TECHNICAL INFORMATION BULLETIN83-7
MIXED MODEOtt .FOR GROUP 4 FACSIMILE
SYSTEMS
NOVEMBER 1983 DTICELECTE
SFEB 6 1984[
D
APPROVED FOR PUBLIC RELEASEDISTRIBUTION UNLIMITED
OIC FILE COPY 84 02 06 ,048"1 ' ',,".'-'.':9,"2- *.'- .- : ... ' " .-". .''.. ", .'''',,. ", " ..'.','3,
7 7- 70 72 71:1 7
REPORT DOCUMENTATION PAGE READ INSTRUCTIONS_______________________________________ BEFORE COMPLETING FORM
1~ REPOT GOVTE ACCESSION NO. 3. RECIPIENT'S CATALOG NUMBER
NCS-TIB-83-7/3' k_______ _____
C. TITLE (and Suft5.l) S. TYPE OF REPORT A PERIOD COVERED
Mixed Mode for Group 4 Facsimile Systems _ _ _ _ _ _ IeorG. PERFORMING ORG. REPORT *UMBER '
7. AUTHON(a) 6. CONTRACT OR GRANT NUMBER(*)
Alan Deutermnann DCA100-82-C-0047
6. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT. PROJECT, TASKAREA & WORK UNIT NUMBERS
Delta Information Systems310 Cottman StreetJenkintown, PA 19046______________
11. CONTROLLING OFFICE NAMIE AND ADDRESS 12. REPORT DATE
1 National Coiieunications System N -her 7, 1983Attn: NCS-TS 13. NUMBER Of PAGES
Washinizton. D.C. 20305 122I.MONITORING AGENCY NAME 6 AOORE43(11 0I0baunt frm Corollfing Ottice) IL. SECURITY CLASS. (of Ohio "aPon)
Unclassified
I~.DECL ASS11FICATION/ OWNGRADINGSCHEDULE
14L 15TUBUTIOM STATEMENT (of W Life pov
Approved for Public Release; Distribution Unlimited.
17. OWRIUTION STATEMENAT (of II. abo&ua ifood to Week 20. It Ef110rnt barn Repel)
I&. SUPPLISKINTANY NOTIES
ILSay WE on"S (CoutMe eawrae side so eey~u MW Aftudp Are o nmber)
Mixed Mode, Facsimile, Group 4, Segmentation
SL AUSTRACr(WIu meww Na inaiemV m I*U5 Of Week smog*)
The purpose of this of fort in to invest igate the more advanced Mixed Mode codingtechnique which will be one service of Group 4 facsimile, In a mixed mode systthe Information printed on a page Is divided into two parts - symbols (letters,numerals, punctuation, ae.) and graphics (logos, signatures, sketches, etc.This study examines possible techniques for segmting a document Into graphicsod symbol -areas, and assemble a code that represents the entire doctinent.Four segmntation techniques were selected for analysis as followas SYMUOLIUOVAL/SCAN LMl; SYKUOL R3WA~L/LInI OF SYMUOLS; NXTENDID TILETUX; STMEDL -
00W Or I MW so i UNCLASSIFIE&Dt
StftITt" CLASS FIATION OF T"IS PA49 ebMm ose. Wo0
* 3 REMOVAL/HYBRID. These techniques were designed to differ from each other asmuch as possible, so as to display a wide variety of characteristics. Foreach technique, many minor modifications would be possible, but it is notexpected that these modifications would alter the conclusions drawn from the
Accession For
NTTS CRA&InjTTC TAB
*
Unannounced EIJustificatio
Distribution/
Availability CodesAvail and/or
% % %~
I Z~fTI ..... ..... .-
NCS TECHNICAL INFORMATION BULLETIN 83-7
MIXED MODE FOR GROUP 4FACSIMILE SYSTEMS
NOVEMBER 1983
PROJECT OFFICER APPROVED FOR PUBLICATION:
DENNIS BODSON EDWARD P. GREENESenior Electronics Engineer Acting Assistant ManagerOffice of NCS Technology Office of NCS Technologyand Standards and Standards
FOREWORD
Among the responsibilities assigned to the Office of the Manager, NationalCommunications System, is the management of the Federal TelecommunicationStandards Program. Under this program, the NCS, with the assistance of theFederal Telecommunication Standards Committee identifies, develops, andcoordinates proposed Federal Standards which either contribute to the inter-operability of functionally similar Federal telecommunication systems or to theachievement of a compatible and efficient interface between computer andtelecommunication systems. In developing and coordinating these standards, aconsiderable amount of effort is expended in initiating and pursuing jointstandards development efforts with appropriate technical committees of theElectronic Industries Association, the American National Standards institute,the International Organization for Standardization, and the InternationalTelegraph and Telephone Consultative Committee of the InternationalTelecommunication Union. This Technical Information Bulletin presents anoverview of an effort which is contributing to the development of compatibleFederal, national, and international standards in the area of digital facsimilestandards. It has been prepared to inform interested Federal activities of theprogress of these efforts. Any comments, inputs or statements of requirementswhich could assist in the advancement of this work are welcome and should beaddressed to:
Office of the ManagerNational Communications SystemsATTN: NCS-TSWashington, DC 20305(202)6992-2124
DELTA INFORMATION SYSTEMS, INC." 310 COTTMAN STREET JE4KINTOWN, PA 19046
(215) 572-540
.J.
MIXED MODE
FOR GROUP 4 FACSIMILE SYSTEMS
November 7, 1983
Final Report
Submitted to:
NATIONAL COMMUNICATIONS SYSTEM
Office of Technology and Standards
Washington, D.C. 20305
Contracting Agency:
DEFENSE COMMUNICATIONS AGENCY
Contract Number
DCA100-83-C-0047
§ubmitted by:
D E L T A I N F 0 R M A T I 0 N S Y S T E M S, N C.
V "N N. * ~ .. *j
Table of Contents
1.0 Introduction 1-1
2.0 Develop Candidate Mixed Mode Algorithm 2-1
2.1 Symbol Removal/Scan Line 2-1
2.3 Symbol Removal/Line of Symbols 2-2
2.3 Extended Teletex-CR/LF op-tion 2-3
2.4 Symbol Removal/Hybrid 2-4
3.0 Task 2 - Measurement of Compression 3-1
3.1 Methodology for Measuring Comp. 3-1
3.2 Assumptions 3-13.3 Calculating Compression 3-3
3.4 Scanned Document - CCITT No. 1 3-6
3.5 Computer Generated Documents 3-25
4.0 Task 3 - Analysis of Results 4-1
4.1 Compression 4-1
4.2 Complexity of Implementation 4-3
4.3 Commonality 4-4
5.0 Conclusion & Recommendations 5-1
Appendix A: CCITT Draft Pecommendation S.a
Appendix B: Combined Symbol Matching Facsimile Data CompressionSystem
.- - 7. W. . 7
1.B Introduction
This document summarizes work performed by Delta Information
Systems, Inc. for the Office of Technology and Standards of the
National Communications System, an organization of the U. S.
Government, under contract Number DCAl0-83-C-0047. The Office of
Technology and Standards, headed by National Communications System
Assistant Manager Marshall L. Cain, is responsible for the
management of the Federal Telecommunications Standards Program,
which develops telecommunication standards whose use is mandatory
by all Federal agencies.
Consideration is now being given to possible CCITT standards
for Group 4 Facsimile which refers to the transmission of an A4
sized page over data networks containing error control. It is
likely that the basic coding technique for Group 4 transmissions
will be some advanced form of the Modified READ code, which is the
optional compression algorithm for Group 3. The purpose of this
study is to investigate the more advanced Mixed Mode coding
technique which will be one service of Group 4 as shown in Figure
1-1. In a mixed mode system the information printed on a page is
divided into two parts - symbols (letters, numerals, punctuation,
etc.) and graphics (logos, signatures, sketches, etc.) The purpose
of this study was to examine possible techniques for segmenting a
document into graphic and symbol areas, and assemble a code that
represents the entire document.
1-1
UL LJ LLJ 000W. . I-Li LUJ I-
C/,o- CDI-C
SUj C,, Ci)
C ~ ~ LL - 0LUL4L >- >0 0o 0
CN W 000P4 )LL L
C) wCu c) --r
LLLU
LUL
0 L LU C
5%~C LU I-LLI C==
LLJ~ 0D C) L3C LD L
NL CLJ L
LL C C/)
LLILUJUi - I/ LLJ - _ _
-iIL CLJ...JL.
L LL C) LU>.
C,,II wLLI Co
f-3 l)
ui1-2
V.~~~C Lo- 4.... -*'~ '1 \~' ' %Cu
Parameters to be considered include compression, commonality with
facsimile and TELETEX 1/ transmissions, and complexity of
implementation.
Four segmentation techniques were selected for analysis. The
techniques were designed to differ from each other as much as
possible, so as to display a wide variety of characteristics. For
each technique, many minor modifications would be possible, but it
is not expected that these modifications would alter the
conclusions drawn from the study.
The segmentation techniques analyzed are:
- SYMBOL REMOVAL/SCAN LINE
- SYMBOL REMOVAL/LINE OF SYMBOLS
- EXTENDED TELETEX
- SYMBOL REMOVAL/HYBRID
Section 2 presents descriptions of the four mixed-mode
segmentation alternatives considered. Section 3 describes the
assumptions and methodology for measuring compression, and the
compression computations themselves. Section 4 discusses the
commonality of each alternative with Group 3 facsimile, Group 4
facsimile, and TELETEX. It also summarizes compression and
discusses the complexity of implementation of each technique.
Finally Section 5 compares the alternatives and draws conclusions.
The CCITT has determined that the 7 layer OSI (Open System
1/ TELETEX refers to a CCITT recommendation which is now underdyvelopment for oommunication between word processors.
1-3
% r*'..** . , . , ; -s.. -- -,.-., -,.. -.--.. ,... .- -, -- ,, , , , 5 4 , 4 4 4 %, .%." '. ' .% .' . ,-. - " . ,". .. " ", . ."
Interconnect) protocol which has been developed by the ISO
(International Standards Organization) will be used for Group 4
facsimile. Figure 1-2 illustrates the top 4 OSI levels
emphasizing the relationship between the Teletex and Group 4
facsimile services. The S. and T. are the designations of the
CCITT Recommendations for each protocol layer. Note that S.a is
the key recommendation for mixed mode operation. This standard
has not yet been finalized. The most recent draft of this
recommendation is included in Appendix A for reference purposes.
1-4
* *-.- >..ft.n, 76 . - V
ft-io PTELETEX IGROUP4
___._TELETEXFACSIMILE
6 T IAPPLICATION:': 'LAYER
:-TERMINAL TERMINALLaS.60 TEIa LEVEL 7
" EQUIPMENT EQUIPMENT
CHARACTER I FACSIMILEREPETOIRE, T,b CODING/
S.61 CODING, I CONTROLCONTROL
" I _PRE-MSENTATION
IMODE . ILAYERLEVEL
PRESENTATION CONTROL PROCEDURES FOR 6TTX S.a TELEMATIC SERVICES
MIXED MODE: DOCUMENT STRUCTURE/ PRESENTATION/ CONTROL
S.62 ESTABLISHMENT & RELEASE, TOKEN MANAGEMENT ISSTRUCTURE INTO DIALOG UNITS, SYNC & RESYNC ISESSIONSIMPLEX, HALF DUPLEX, DUPLEX LAYER
5
(S.70 END-TO-END TRANSPORT 1TRANSPORTSEGMENTING ILAYER
,LEVELi 4
FRAMEWORK OF CCITT RECOMMENDATIONSFOR GROUP 4 FACSIMILE APPARATUS
FIGURE 1-21-5
4'7
4-
2.0 Task 1 evelo Candidate Mixed M-ode Algorithms
Four mixed-mode segmentation techniques are selected for
consideration in this study. The techniques are:
Symbol Removal/Scan Line
Symbol Removal/Line of Symbols
.4 Extended Teletex
Symbol Removal/Hybrid
In the three symbol removal techniques, the black pels
associated with recognized symbols are coded and "removed"
" (changed to white), and then the entire document is encoded using
the Modified READ code, including areas where the symbols were.
In the Extended mELETEX technique, the Modified READ code is used
only for areas that do not have encoded characters. All
techniques presume existence of a stored library of symbols.
2.1 Symbol Removal/Scan Line
This coding technique is very similar to the Combined Symbol
Matching algorithm which is described in Appendix B. In this
approach the document is scanned, from top to bottom, and from
left to right, until a group of black pels is encountered that
matches a symbol in the stored library. All black pels within the
rectangular symbol space are then changed to white, and the symbol
code and position are recorded. After the symbols have been
"removed", the documnent is rescanned in principle and encoded
2-1
using the Modified READ code (k-0, no EOL code). The detected
symbol codes are inserted before the READ code of the scan line in
which the top of the symbol occurs. The presence of a symbol code
rather than a READ code, is indicated by a single bit at the
beginning of every scan line. If the bit indicates that there are
symbols on the scan line, the 8-bit symbol code follows, and this
in turn is followed by an 11-bit horizontal position code word,
(211=2,048, being greater than the 1,728 pels in the scan line).
This may be followed by additional symbol/horizontal-position code
pairs for any other symbols that may have been detected on the
scan line (in order of horizontal position). Finally, the symbol
data is terminated by a special 8-bit symbol code that indicates
there are no more symbols on the scan line. Then the READ code
for that line is transmitted.
Notice that in this technique the recognized symbols will be4-'
encoded as they are first encountered by the scanning process,
regardless of where they appear relative to other symbols or
graphics. The vertical position of the symbols is implied by the
scan line on which the symbol code appears.
. 2.2 Symbol Removal/Line of Symbols
In this technique, as in other symbol removal approaches, the
4symbols are detected, "removed", and their codes and positions are
4recorded. The symbols are then organized into lines of symbols,
based on symbol position, height, hang down, etc. Account is
taken of small amounts of line skew, and a single vertical
position is assigned to the entire line of symbols. When this
2-2
.W N
-1: . Ia RR- .- - - -4.
process is complete, each printed line on the document should be
contained within a line of symbols. Spaces between symbols having
several different widths up to about 2 normal symbol spaces, are
filled with appropriate blank characters. If the space between
symbols is greater than 2 symbol spaces, the line of symbols is
broken into segments.
The entire document, less recognized symbols, is transmitted
using Modified READ code. When a scan line having the vertical
position of a line of recognized symbols is encountered, a special
12-bit code (which could be an EOL code) is inserted. This
changes the mode from graphics to symbols. This is followed by an
11-bit code giving the horizontal position of the first symbol.
Then the symbol codes for each symbol in the segment are sent,
followed by a special 8-bit end-of-segment symbol code. This, in
turn, is followed by an 11-bit distance-to-the-next- segment
symbol code. The last segment of symbols on the line is followed
by a special 8-bit end-of-line symbol code instead of the
end-of-segment code. This changes the mode back to graphics, and4
Modified READ code is continued until another scan line with a
line of symbols is encountered.
As with the other symbol removal techniques, a recognized
symbol will be encoded wherever it is located, since lines of
symbols may overlap vertically, and each line of symbols may
contain as few as one symbol. There may be some inaccuracies in
positioning symbols, since spaces between symbols of 1 or 2 pels
will probably not be encoded, and the horizontal position of a
symbol code could be in error at the end of a long line of
2-3
- tX I
symbols.
2.3 Extended TELETEX -R/LF option
In this approach the entire document is divided into
character spaces, except for areas that are defined as graphics,
as discussed below. All character symbols, including blanks, are
transmitted using 8-bit symbol codes.
The graphics are transmitted by Modified READ code as they
occur within a line of symbols. First, a special 8-bit symbol
code is used to designate the transition from symbol codes to
graphics. This is followed by an 11-bit code giving the width of
the graphics area. (The height of the graphics area is defined by
the height of the symbol font.) Then the READ code for the
graphics is sent. The length of the READ code is defined by the
width and height of the graphics area, so the transition back to
symbol codes does not require a code.
In the CR/LF option, instead of transmitting a series of
blank symbol codes at the right of the line, a special 8-bit code
is used to designate the last-symbol-on-the-line. This, of
course, would have to be to the right of any graphics on the line.
This last-symbol-on-the-line code would direct the receiver to
start on the next line of symbols, and would replace the CR and LF
codes of TELETEX. For reasons of commonality it may be preferable
to keep the two standard TELETEX symbols for this purpose.
This technique is considered primarily as a method to
incorporate graphics (such as logos and signatures), into
2-4
computer-generated text. Therefore graphics areas are defined,
probably by the user, as rectangular areas which may contain a
mixture of graphics and symbols.
Since all lines of symbols must have proper spacing, symbols
that are not aligned with the majority of the symbols must be
treated as graphics.
2.4 Symbol Removal/Hybrid
This technique combines features of the other two symbol
removal techniques to make it more robust then either in that it
is designed to handle both isolated (or arbitrarily located
symbols) and symbol strings in lines or segments.
In this technique, as in other symbol removal approaches, the
*symbols are detected, "removed", and their codes and positions are
recorded. Spaces between symbols (up to 2) are filled with
appropriate blank characters.
The presence or absence of a symbol code, rather than a READ
code, is indicated by a single bit at the beginning of every scan
line. In addition, a single bit preceding each symbol code
indicates whether the symbol is contiguous or not, i.e., not
followed by more than 2 blank spaces. If the symbol is not
contiguous, it is preceded by a horizontal position code otherwise
the symbol code follows immediately.
A special 8-bit symbol code terminates the symbol string at
the end of the line of symbols. Then the READ code for that line
.is transmitted.
2-5
4I
3.0 Task 2 - Measurement of Compression
3.1 Methodology for Measuring Compression
For each of four proposed mixed mode techniques, an estimate
of compression has been made. Extimates of compression were made
for CCITT test Document 1 and for two computer generated
documents.
It should be recognized that compression values calculated in
this report are estimates only, and should not be regarded as
actual measured numbers. However, it is expected that the
relative compressions of the various segmentation techniques are
accurate, since the same assumptions were used for all of them.
3.2 Assumptions
In making compression estimates, the following assumptions
were made:
(1) Each symbol is encoded using 8 bits, which allows
up to 256 different symbols.
(2) Several of the 256 symbol codes can be made
available for indicating termination of symbol
transmission, or other requirements of the segmentation
technique employed.
3-1
(3) A stored library, suitable for the document being
transmitted, is available at both sending and receivingterminals.
(4) Bits requirel to identify the proper symbol library
to the receiving terminal are neglected.
(5) The stored library will accommodate either fixed or
proportionally spaced fonts, including several widths
for word spaces.
(6) All characters of the principal font used in the
document are in fact recognized as such, and will be
encoded as symbols, subject to the rules of the
propose6 technique.
(7) Lines of symbols can be accommodated despite slight
skews of the printed lines.
(8) The characters of the principal font include math
symbols, italics, and Greek letters, but not subscripts
or superscripts, or long horizontal or vertical lines.
(9) Graphic data is transmitted using the modified READ
code, without EOL's and with k-C.
(10) The number of bits required to transmit increased
width of white spaces by means of Modified READ can be
neglected. This follows because the spacing between
groups of black pels (such as symbols) usually only has
3-2
. . .. . . . . . . . . . .~
-to be specified once, and the READ code length does not
grow rapidly with the length of a white run.
(11) Each A4 source document normally 216 x 297MM,
*consists of 2,376 rows with 1,728 pels per row (ie,
resolution equals 8 pels per mm, or approximately 200
pels per inch).
(12) .Scanned documents, stored on tape, are retrieved
onto disk as an image file commensurate with a 16-bit
computer word size. Thus the computer image file
consists of 2,336 rows of 1,728 bits each row. Since
the computer image contains all the black pels of the
original document, any process, such as the modified
READ algorithm, performed on the computer image may
represent the same process performed on the original
document with negligible error or vice versa.
(13) Code transmissions will not experience any
transmission errors, so addition of redundancy for
error control is not required.
3.3 Calculating Compression
For each technique the number of bits required to construct
the message is totaled. This includes any flags, tag bits, symbol
codes, end of symbols on line, end of segment graphics, symbol
mode changes and horizontal and vertical positions. The number of
bits required for each of these functions is given in Table 3-1:
3-3
MW . * 4'
Table 3-1
Table of Bit Requirement for Data Function
Data Function Data Requirement (Bits)
Scan Line Flag 1
Contiguous Symbol Flag 1
Symbol Code 8
End of Symbols on Line or Segment 8
Symbol to Graphics
Vertical Position-Symbols on Line 12
Graphics to Symbols
Horizontal Position 11
The compression is calculated by dividing the total mesage bits
into the total number of pels, as referred to the source document,
which is always 2,376 z 1,728 a 4,195,728.
In the three symbol removal techniques, the black pels
associated with recognized symbols are coded and "removed"
(changed to white), and then the entire residual document is
encoded using the Modified READ code (k-, no EOL code),
including the areas formerly occupied by the "removed" symbols.
In the Extended TELITEX technique, the Modified READ code is used
only for areas that do not have encoded characters.
Values for document related parameters used in calculating
compression estimates are given in Table 3-2.
3-4
- * ~~ .' -'o ,'' .',L"___ __CY ,; :C&.L-A. , .- ,
AA
'.P4
0
r4
Q0 ('4 N ~ ' -H co aw ch r-4 -
0 ri u-Ihia
6C1
*0 0 0 r4 . 0 a '
41 0 0n N A WCLI IM 0. x- I rI
M- V-4 0 0s 04 . 4 C t
AI o r-IO % 4 E4 s
:i C ON- 41 0' ' -
04 L0 u ) I
*1 x 4 to "
Ai P40 84# V C1
a 0 " a 0>$ A 01 w
a 0 1 o 62 0I
01 041c w C Oc CI C"416a'O VC 0 04 ".4 "'4 4V
CO"41 0'. X'6 0CK. -OEM C- 0 41 H' MV 4 C 44 S
'40 14C C 44 61E s 0- 039 1 0 39 61 to0 U -4161
100 041 5x go 0 to CC 00 61"m 41 C Ai4 0- ta u cl.
N-4 0.6 61 ". to "4 X U0 0. 610 4 0 040 to qUK
0 61 to~6 C e -P4 "4 .4 00
,.0 C46 0.6 c OV-4 41 61 .0
61 0 0 41 U 6C10. OCx. * 4~ 4duI
~' ~ 0.304
4.7
3.4 Scanned Document - CCITT No. 1, Figure 3-1
3.4.1 Symbol Removal/Scan Line
Figure 3-2 illustrates the composition of a mixed-mode
message using this technique. As indicated in Figure 3-3, all of
the typewritten symbols are recognized, encoded and "removed".
The presence or absence of a symbol code, rather than a READ code,
is indicated by a single bit at the beginning of every scan line.
A special 8 bit symbol code terminates the symbols on the scan
line and indicates the end of symbols and start of graphics. Each
symbol's horizontal position is independently encoded. Since the
vertical position is implied by the scan line on which the symbol
code appears, no account need be taken of spaces between words or
between line segments. The Modified READ code (k-sb, no EOL code)
is applied to the residue (Figure 3-4) after removal of
typewritten symbols. A summary of the compression estimate for
the Symbol Removal/Scan Line Technique applied to the CCITT-I
document is presented in Table 3-3.
3.4.2 Symbol Removal/Line of Symbols
Figure 3-5 illustrates the composition of a mixed-mode
message using this technique. As indicated in Figure 3-6 all
typewritten symbols are recognized, encoded and "removed". The
symbols are organized into lines of symbols with spaces between
symbols (up to two) filled by blank characters. The resulting
string of symbol codes is preceded by a graphics-to-symbols code,
a horizontal- position-of -the-first-symbol code and is terminated
3-6
THE SLEREX COMPANY LEW=TESOAPMS LAWIE mac SOM DOSSUT.- 01H 5 9 U
same (96 13) $140 . 1..m IS
Out Not. 350/PJC/IAC Loth January, 1972.
or. P.m. Cumall,Yliaing Survey* Ltd..
leading.lefts.
Dear Pete,
Perit me to introduce you to the facility of facsiiletranmission.
to facaimnile a phototall is caused to perform a fater &ca ovethe subject Copy. The variagioma of print density on the documntcausne the pbotece 11 to enerate an anlgous electrical video sismal.a" saignal is me" to MAdUL"* a Crier, ukich ia trossm.tted to artint deetistiom ever a radio or cab1e comuications lik.
At the reenta terminal,* dwooduLation recomatructs the videoGipal. Which is tmed to andelate the density of print produced by aprinting device. This device is scanning in a raeter seem ayuchrouiaedwith that at the trmasting terminal. As a result. a fasim~saom of the subject documnt is preduced.
Probably you hae maoe for thia facility in your .rgaoiaation.
Yours aincerely.
P.s. cams* Group Loader - facsimile Reearch
*m~we., do VM L~tOt&M
Figure 3-1CCITT Document Number-1
3-7
Figure 3-2'.5
MESAGE COMPOSITION
SYMBOL RMOUVAL/SCAN LINE
Scan SYNLine PEIES
201 0G
202 F,
203 0G
204 0 G2"5 11 S HPOS BOS G
,26 i0 G
207 1 S POS S I S S HPOS )S G
208 0 G
209 0 G
210 0 G211 1 S 0S S HPOS ~ G
212 0 G
SIM PIES 1 indicates at least one symbol on scan line - 1 bitG graphics mode using Modified READ code - variable bits
1 S symbol code - 8 bitsHPOS horizcntal. position of symbol - 11 bits.;6 end of symbols on scan line - 8 bits
.3-
3-8
* .
*A.
N.. THE SLEREXE COMPANY Li ED
SAFORS 301. LAB .BOL DORSET - IBM 25 £BEuromn -- s (96l LI) )II7 . lu t214h6
Our ftf. 350PJCEAC 18th January, 1972.
Dr. P.M. Cundail.Mains Surveys Ltd.,N royd Read.PIS ins,
Ber Fae.
Perit - to incructs you o Cho ftcilicy of facsimile-* ela. .. KImmu&iO 01
to facsimile a piotoc.Ll to caused to Perform a raster scan overthe subject copy. The variaiaons of print density on the dlocincCAM" the !Loocell to inemerace el mzlou eloctriat video si al.[his staal is se4 o nnulaNe can'ier which La cransmicted to areme. dasttatoe over a radio or ae cOlmicatioua Link.
At the retme ttrminaL dindnlaejae raconsruca the vidaoaisal. which is Used to Muaz ch d ity of pri.nt produced hr aprnting device. This device is scamn in in a ater scan synchrouesiedirth a at t hre rensmcctia terminal. As a result. a facsimile
cowof the subiect decampst i r oduced.
Probably you hawe uses for this facility in your orsanisation.
Tours sincerel 7
• PF.J. UlOS
Figure 3-3i Symbol Removal/Scan LineCCITT Document Number 1
3-9
I_% m l" ' I. " :''" " " -. ,,.- '.- .,,.-....'..'..•. .,-.
11-43 A04 TUE.. 19I OCT. 19!3#
LO,,1 C CLP' STARTING ATp PEL I APPQ#0~ RECOPL' LENGTH 17'I
I0
THE SLERE-XE COMPANY LImYrmSAKM. LANZ . 9OUS . WaM E. a Z a n
(06ni 0 ) S167 1m 3M
.
-'"
O0. "
.,-o
Residu afte remova of symbols11
'.
%ai°%
.;...
".o
* Figure 3-4
~Residue after removal of symbolsCCITT Document Number 1.
i 3-10
4 " 4b %
4...)
t.
0 0
I" N
41-j
r 4 0 .
.0 (j(E-, ,%- , - 0
0~ '4 01 04 .'La
0 : I
>1 r- .4 0
eU *14 1 -4"4 61 w1 C
0", UI 0 0
a 12 ga- (00 to rd ro
01 01 4
0 '-.0
(D 0
00
0W 0
3-11 J
Figure 3-5
MESSAGE COMPOSITION
SYMBOL REOVAL/LINE OF SYMBLS
Scan Symbol
Line Line201 G -
202
203 G
204 G
205206 1 G/S HPOS E S S S S OLG
207 G
208 G
209 G
210 G
211 2 G/S IIHPOS s S OSIDIST S SI G
212 G
G graphics mode using mdified READ code - variable bits4S symbo1de - 8bits
G/S indicates change fron graphics to symbols - 12 bitsHPOS horizontal position of first symbol in line - U bitsBOL end of symbols on line - 8 bitsBOS end of symbols cn segment - 8 bitsDIST distance between segments - 11 bits
~3-12
5N.% X
-~~ 9% 74 TP. S--
THE SLEREXE COMPANY LI1WYTE
uas Wma 90 U) OSU - UIM"I
bu W. 3~W551.-
F.4~4 It. C -"
Ir Pad
lp;- w to antrodwA po to the faecilit of foosuil
lie facsimea obs0 ha s amused to etaov a raster *can oustit ubject copy Th *Vatat as of vizat ddMi a M"At
=the pbCeauI to 2"ate a am- 1 electrialk vdoLVamgSI. I4 L4 ed to induiece 4 CAMTter, VkL@& La tiaaantd toa
ine .089100sovr a raio or COLO cmmacaioea iiaU.
I~ thrae tos oal. domoti rsecomtr~asteWde
rias devie. IbTis taec ;I seamata j& a tutor! Samc *haaed isx bea at the craneantcia tWmiAa. As a teouL t. ; taca~
0the uject document is ttducod4
Pfrobabiy ovn have uses for Cif fiii LA Taut or ma465ea-.
4I Yours aimeeralu
~~inseu a&.. am
Figure 3-6Symbol Removal/Line of Symbols
CCITT Document Number 1
3-13
with an end-of- symbols-on-line cod . More than two spaces
between symbols on a line breaks the line into segments which
except for the last segment are each followed by an
end-of-symbols-on-segment code and a distance- between-segments
code. The last segment in a line is terminated with an
end-of-symbols-on-line code. The Modified READ code (k-wP, no EOL4°'
code) is applied to the residue Figure 3-4. A summary of the
compression estimate for the Symbol Removal/Line of Symbols
Technique applied to the CCITT-l document is presented in Table
3-4.
3.4.3 Extended TELETEX - CR/LF Option---43-4
. Figure 3-7 illustrates the composition of a mixed-mode
message using this technique. Figure 3-8 shows symbols encoded by
the Extended TELETEX technique (with transmission of CR/LF symbol)
and also shows boxed-in areas of Graphics transmitted by the
Modified Read code (k=-D, no EOL code). All typewritten symbols
are encoded as are blank characters used to space over: to the
first character on each line of text and to the left hand
'4 horizontal position of the boxed-in areas of Graphics (see figures
3-9 through 3-14). Symbols-to-Graphics and Graphics Width codes
guide deployment of the Modified Read Code. A summary of the
compression estimate for the Extended TELETEX-CR/LF Option
technique applied to the CCITT-l document is presented in Table
3-5.
* -Y 3-14
a C'4 co -1 qW ef anAj r- -4 co m'~ ft
a 4b
% a%
0 0I.4 .IJ M "4 ~
P- 0J oC
a 44o C o*61 0d -'4E -4 Q - 40 > 0
.0 041 I o 954L 0 (P V "
4 o tp c ON X w0 0 UA U %4 - on02 1 01 C 0 v)
'0 r-4 U 0 Mid 0 0 41 to a)
c 0- 1 Of+n 0 u W4
4 44 C 44M
41 '0 4.PC to 41
"4 W a)
A 3-15
Figure 3-7
MESSAGE COMPOSITICN
EXENE TELETEX
-' SymbolLine
L1* 5 S S S CR/LF
6 S S S S S S S S S CR/LF
7 S S S S/G Width G CR/LF
8 S S S S S S S CR/LF
9G Widthic moI Isn Moife R~ I od - a I " t
10 S S S S S S SISIaL
U - S S S S S S S l v l- F ;q12 S/G Width I G IS IS/G IWidth GS
*13 I I a F
LEEN
G graphics mode using Mdified READ code -variable bitsSVq2 symbol code - 8 bitsCR/LF carriage return/line feed - 8 bitsS/G indicates change from symbols to graphics - 8 bitsWidth width of graphics - 11 bits
.313-16
1U.pnsi Mae (94S 12) 91617 Maxm LLUS
Ou e.3* /JI~ th Janur.171
I .m. &mslMinin StarvQ Lt4.,
2044
*e to Lacroduce you to ct facility of facsiLlei
a5.9. on a.1 heec paue o r -ru raster scan evert e ac c * va~a~ea Print del!sity On Ste docuinc
.aC
nThs2=1 ievice. Ca5 udulace as carriiar ufic a s ratrananhia
At he an. t al Aarsutazcea
cop of the subject docuio is preducedj
Probably you have uses for tnzis fai-t Lfi !our organisatcan.1
Yours sincerely,
.4I
.. 4- m s
Gru odr?-sa4 oerh
3.1
.V.
|.- ",.
*i
12:28 PM MON.. 21 NOV.. 1983
PLOT (LOGO > STARTING AT PEL * 1 (APPROX.) - RECORD LENGTH Be
LINES READ , 80.
-, Figure 3-9Logo Graphic
CCITT Document Number 1
3-18
.. ,n y u . q , q, !u , . . . .mp• J l
5t,
4 -I
aI
12:27 PM MON., 21 NOV., 1983PLOT <SLRX > STARTING AT PEL I 1 (APPROX.) - RECORD LENGTH 848
, COMPANY LIMED
LINES READ - 44.
Figure 3-10
SLEREXE Graphic
CCITT Document Number 1l,
3-19
- r - . "b "I% .%. - . .
12:26 PN MON., 21 NOV.. 1983PLOT (SAPR > STARTING AT PEL I 1 (APPROX.) - RECORD LENGTH 624
SAPORS LANE. BOOL . DORSET - BK 25 8 ER
, LINES READ * 35.
Figure 3-11
1. . SAPORS LANE Graphic
CCITT Document Number 1
3-20
'/4, ' ,' ' ', ' ' '. . " . .• " . " . " ." , '""""." . ' " . " . ' % % . . ,
*e,
12:25 PMl MON., 21 NOV.. 1983PLOT <PHON > STARTING AT PEL I (APPROX.) -RECORD I-ENGTH 544
,m~m m (945 13) 171' - 'nx135
" LINlES READ ,, 3J.
J Figure 3-12
i TELEPHONE Graphic
P--' ,CCITT Document Number 1
3-21
-w. . ... . .-aV ",% % % ._
a>."
12:35 PM NON., 21 NOV., 1983PLOT <SIGN > STARTING AT PEL E I (APPROX.) - RECORD LENGTH 192
.4
.4
LINES READ , 88.
Figure 3-13
Signature Graphic
CCITT Document Number 1
3-2
3-22
.4.'% -, . ' . ' . ' ,. . . .,, ' . ' - . ' " . . . , . . ' .- . . - - . ., " - , ' % - - ' .
12112 PM MON., 21 NOV., 1993PLOT <REGIS ) STARTING AT PEL I (APPROX.) -RECORD LENGTH 565
Re isteed in *fald No.17036&Mstayed 0mO.: 0O Vicaft La. Ilford. Beoi.
LINES READ * 43.
Figure 3-14
Registration Graphic
CCITT Document Number 1
3-23
C4 %a C4 0 %0
C4 qw
0
4' N C. (4 0 0Ch ,4 O -I V-4 I-
a L)In -I
00 r: mco0 Ch
r.' E-
.13 C4Go 02 r- ) L
4' 4 0 V
.0 to u )4.2 03 (D m
0 v u410Va) 0-gto 4
0VI 0 'g0
go -r2 W2 -. U
V 2 0 r Iuo co E) u 0U w4 0 w 02
.0 UC0 C
.4>t .54 024.~to U 0 L 4 m
%~4. +4'3V24
3.4.4 Symbol Removal/Hybrid
Figure 3-15 illustrates the composition of a mixed-mode4,.%
message using this technique. As shown in Figure 3-16 all
typewritten symbols, including spaces between symbols (up to two),
are recognized, encoded and "removed". The presence or absence of
a symbol code, rather than a READ code, is indicated by a single
bit at the beginning of every scan line. In addition a single bit'
preceding each symbol code indicates whether the symbol is
contiguous or not. If the symbol is not contiguous it is preceded
.* by a horizontal position code otherwise the symbol code follows
immediately. A special 8 bit symbol code terminates the symbol
string at the end of the line of symbols. The Modified READ code
(kaw, no EOL code) is applied to the residue Figure 3-4. A
summary of the compression estimatefor the Symbol Removal/Hybrid
technique applied to the CCITT-I document is presented in Table
3-6.
3.5 Computer Generated Documents - A and B
Computer generated documents A and B are shown respectively
in Figures 3-17 and 3-18. Compression estimates for the computer
generated documents are calculated in the same fashion as
previously described in Section 3.4 for the CCITT document number
1. In the Symbol Removal/Scan Line Technique symbols are removed
as illustrated: Document A, Figure 3-19; Document B, Figure 3-21.
After symbol removal READ Code is applied to the residue: Document
A, Figure 3-20; Document B, Figure 3-22.
4. 3-25
V
Figure 3-15
Message Composition
Symbol Removal/Hybrid
Scan SymLine Pres
201 0 0- I I
202 _0 0G
.4
203 0 0G
20 00 G
205 1 0 HPOS S _11 S 1 1 S- 1 GO
210 0 0 G
Legend
..4
Sym Pres 1 indicates at least one symbol on scan line 1 bitG graphics mode using Modified READ code -variable bitsS symbol codeR/ -8bits
SPOS horizontal position of symbol - 11 bitsEOS end of symbols on scan line - 8 bits
3-26
S' V
_, 01 0
THE SLEREXE COMPANY LEWiED4SAPOU LAHR - 000A - aLSET. WK 35 AE
%CksF- (a~~z rprAWguiDE
.*1or ZahT 3~~j7 LatrwoJaZU72t--
t~m asto srjt~5 LW...a he faci Lety of facaielI
La earnd To £! . wna a te cm oe
(EtrJ~i t'f, h%;Vari~uo - IQ 7 iaiyo - - dw rwaiItASA toi-~tri trz aeaaiao" lctia 19
!LifJj..Ubic 'ia ued to asd"aisc ri._eei ofnito to~ p.00 1
I abi on Over~ Lltiorh Tcoemqu La m~r~~~t~.
liXt sicaeiO
'a..~rcW"sdt oulc h"6 r~~ili
Leda _q SZ iiniM Euee~'ai6 i-vi i cm~-Lnarse cnVcs aa
a,;
-'aM
Figure 3-16* Symbol Removal/Hybrid
CCITT Document Number 1
3-27
co 0
0. c m - nC444
04.to1>
4o1C
r 41 "4 c W 14 a#ogol~ ~ 0' ('4
41 "4%
-o a Ci
0. 0coi 0 (a5
5*c C 1 4 f-4
.At 1 0 iC cU 0 N.- %00%. to
0 04 Aj 4.
+ 41 .0C2 41 w C-4 cc
0 F - 0 UU 41 Ci) W
ui 0. a ca U Fa
44 41
S4,j0 c,
w 41 0
3-28
*~~~~~r -- a . -71- - *-
Income is an inflow of assets, but it must be recognized
that there are inflows of assets which are not income. Obviously
an inflow of capital funds from stockholders is not income to a
corporation, nor should a business regard as income an inflow of
assets which is offset by an increase in liabilities. Income
consists of an inflow of assets in the form of cash receivables,
or other property from customers and clients, and is related to
. . .......... ...... . ..................... .............. ........................ S A L E S ..........: ' ..' ..... .... ............... ..... .. ....... ... .............. .............. .............. ............ .....: ....... ................ .... .. . ...... ' ... .... .... 7 ... .. ..... . .............. .......... .. . .. .... . ..... C O S T S
t, O I .... .......... ...... ..... .. ........ ... ...... ............ ....... ...... . . . . . . . . ."
7 0 " 0 41 .. . .. . .. . . . . .. . . .. . . .. . . ...- ... ... ... . ........... . .. .... .. . .. . .. . . .
. " G a s . . .. i ...... . .. .:..... ..... . . ... .. . . .... ... . .. . ... .... .... .. . . .. .. . . . ...... . . . .
406, 0. .. ... ......... .e: ... '............. ...... ... ........ ......... ... P.O.T
,." M u ...... ... . . . . . .
1%k? 1$60 IWO 1170 I971 t$72 1973 19374 ,5 1976 197,
" IrI'PHJC!Z OUTPUT
i the disposal of goods and the rendering of services. If income
is earned by selling goods, it may also be called profit; the
I term profit is not properly applied to income derived from the
rendering of services.
A basic criterion for the determination of the period in
which income may be regarded an earned may be stated as follows:
.,
t,,Income should not be regarded as earned until an asset increment
has been realized, or until its realization is reasonably assured.
Page 1
Figure 3-17Computer Generated Document A
3-29
.o-a-'- - .7. . . - . .
--_ ---. .-- -- -
"J FIGURE 2 EXTERIOR VIEW OF PROPOSED BUILDING
.We :re running out of oil - and natural gas. Whether i'
i''
exactly 30 yars or more makes very litle difference in the longrun. As we begin to drill more deeply into hard-to-reach reserves,
" the supply will become more spotty and more expensive. So startwplnning for oil-gas alternatives.
The best is coal. It's conservatively estimated that we have300 years of coal reserves. However, the cost of mining andtransporting it will grow sharply as demand builds. (Much of the
coal %d11 be'difficult to reach, too.)Should a company convert its boilers to coal-burning from
:,,,,:oil or natural gas-burning? In many casts the answer is yes.
,,1 . 7 / . T. A ,-.,. ' I - ',
66% Coolingi .in u4s
"" >EA&:"'' t" " GE° I E71LJ -'L.
-- I
FIGURE 3 PIPR LAYOUT DIAGRAM
Pace 2
~Figure 3-18Computer Generated Document B
Tte3-30
40 er fca eevs oeetecs fmnn n
income is an inflo~v of assets, but it must be recognized
that there are inflows or assets which are not income. Obviousry
* an inflow of capital funds from stockholders is not income to a
corporation, nor should a business regard as income an inflow of
assets which-is offset by an increase in liabilities. Income
consists of an inflow of assets in the form of cash receivables.
or other property trom customers and clients, and is related to
................ ..-...... ....... ........ ............. A E S
. .................. .. .. ....... ......... ...... .. ...... ....... ... .
. .............. "........ .............................. .. ...-. S '
Save ? . .. ...... . 7 .. 7 ..... 7 .... .... ........... ...... ....
7000 ....... ............
the~~~~ di.oa of... .od ..nd. t.l .....r~ of .sev.. ficm
is9 eane by selling .. o.ds, it.... may als be. called prof.t the......
term p.fi is... not ........... ly. a.le to... incom derve fro the...
aae
thediso Symb golsad h rendericng ofinervcs ficmis arnd ysllig ocumeit as becldprft th
tem roi i ntprpel apledt icoed3-31frm h
V. Z%-Z
.. .. .. .. . .. .. . . . . . . . . . . .. . . . . . .
...... . .........m . ......... S A L E S .!........
.. ... ... ... ... ... ..
lo p ...................4... ... ... ..........
loe. .. .. .......... ...... ....... ........ ........... PR.....O........ TS ' O T8 1 " .. . . . . . . . . . . . ..C.'. . . . . . . . . . .. . . . .. . . .; W . . . .. . .
'0 0 . . . ..-. .. . . . . . . .. .. . . .. .. .. . ..4. .. . . .
6000 ~ ~ I fl HT 5 OUTPUT... ........... .... ...........9" .. . ... ..... ... .... .
.. ... ....... . ...... .. ... ...... ..
Figure.... ...-2....
Reiu after reovl.f.ymol
.. . .. . . ...... .... .. . . . . .. . . .. .. .... .
"7l-.
N
.' . ,"FIGURF 2 EXTERIOR VIEW OF PROPOSED BUILDING
We are cunning out or 'il and natural uas. Whethe
exactly 30 years or more makes very little diference i i on
run. As we bea t2 11il more deeply into hard-to-reach eserves,thesu~ly illbecom more spott arid more exensive. So start
Ll'n9n2n f o f oil-gas alternatives.
The. best s coal. It's conservatively estimated that we ave.
300 ers of coal reserves. However, a h costof mLninnd
trans!Irtinq it will ro! pha ly as demand buls Mch ot the
oa 11 be-ditticu o , too.)
Shol a compn S onverr ts boilers to coal-burning fromoil r rnatural gas-Murninq7 Inmany cases the answer is yes,
*- !l J- o r 'ILI ': !I-R !
S.lann olr oi-a lternaIvest"he.. be;.st isn cal.. I' osraieyetmtdta env
i Vp I*°
6 .€O| n htm~
60 er o olrsevsowvr h costomiingdo and
~COOL. TOWiER
FIGURE 3 PIPR LAYOUTDAR.
coaFigure 3-21
,Symbol Removal/Scan LineDocument B
3-33
.." -." -." . ,,: -.ould a company' 'o'.err. its.boiler.s-to coal-burnin'.from
7 - 'b 7~ -
riri TT ~
if+T." AI FANSl . ~ l
I.. -
So I o MWVSP I r n W
Sw ppVl 0 .hoofing
- * -.
L ~ t TC1OL OWERi"O. A . . .
9
IIS:. I I .1 1SRE11 1 1 1 1 1 1 1 1 "
......... BO-- - - - .. .- ... ..
, -I . ... . .
Sl. r . . ,...., .,I L . ,,.,
".9
." I L.4 ~" '; " COOL TOMCR:
'l:l"EMO? " ,+ r A.-C,.
'."- Resdue fte remval f sybol
Figure3-2
Residue.,,.+",,"...,,;.'." after removal..,,,.. of",.., ,. .symbols +.. •.,..,.....q...., ....• . . .- i ,.++• %•. % o"."+"a
Similarly in the Symbol Removal/Line of Symbols Technique,
symbols are removed as illustrated: Document A, Figure 3-23;
- Document B, Figure 3-24.
The Extended TELETEX-CR/LF Option Technique is applied as
illustrated: Document A, Figure 3-25; Document B, Figure 3-27.
READ code is applied to the boxed-in graphics: Document A, Figure
3-26; Document B, Figure 3-28 and 3-29.
In the Symbol Removal/Hybrid Technique symbols are removed as
illustrated; Document A, Figure 3-30; Document B, Figure 3-31.
The results for the four mixed mode compression techniques
are presented in Tables 3-7 to 3-10 for computer generated
documents A and B.
3-3
U.."
'%
S S
[Income is an inflow of assets, but it Must be recognizedi
that there are intlows of assets which are not income. Obvio-u-s y
Ian inflow of capital funds from stockholders is not income to al
Icorporation, nor should a business regard as income an in ow
-a assets which is offset by an increase in liabilities. Income]
1consists of an inflow of assets in the form of cash receivab es,
Jor other property from customers and clients, and is related tol
.. ..................... S A L E S.a ' ..... . .. I .. .. .
...... . . .. ... ... ... .T.
CO-T
-------- - -:::: -- -- I - - R F T
19. 96 Ilb- 1'. 19. 1.. 2 1r 3 -4 L -7S i-. ;77
' i-fPH1C- OUTPUT
tIe disposal o goods ang the rendering or services. I incom-e
is earned by selling goods, it may also be called profit- thel
Iterm profit is not propcrly applied to income derived from the
Render ing of servicesj
JA basic criterion for the determination of the period i
[which income may be regarded as earned may be stated as follows:
rincome should not be regarded as earned until an asset incremen~t
has been realized, or antil its realization is reasonably assured.
Figure 3-23Symbol Removal/Line of Symbols
Document A
'a 3-36
%a-
--~ ~ - ----- --
----------------------- -------
-n~o2- ool-ads.Wehri'
rwm e .... -SHE
Etspl i~i,1-t
I |Wen aoe runninsa oteof -v;s..5 RV weha
Wean arn atural a.Whetheris
-'; lrun. As we De in to dril moedel noh~-oca cs~e~
%''% the supy will become more SoDotty and more expensive. So satj<'4 ]anning or 0±i- as a ternatives.]
ITne oest s coa t s conservaive estimatd that
ears ot coal reserves. However, the cost of minin@ andtransporting it wil grow sharply as demand builds. (MuC o te
oa W1 bedi icut to reach too. l
!Should a com an convert its boilers to coal-burninK ro,,mI-. or natural cas-burning? In many cases the answer is yes.1
solar oxn I I::i'~m K. pp|e ,G1:- ',,,-,fli I --
* *~ 66. cooing -*J#Cfd
K/ "/ .I|Ir nit' .
- COLD ftESCP'IP-1
COOL TOER
P OT AOILC
WHI C P
%,,
Figure 3-24Symbol Removal/Line of Symbols
Document B
3-37
a,-
that theres ar inlw of ases hc ae Lt noe bio
leoan nlwo aia ud rmsokodr snticm oa
coprton o hol usns eue sinoea ifo f
asstswhchis ffetbyan nceae n labliie. Inomerjc~w o nssso nifo fast ntefr fcs eevbe.
or ~ ~ ~ ~ ~ ~ ~~ R othTS jrpryfo utmr n letadi eae o
is ea ne ... s. .. in ...... .. .. a. ... s .... ec o c it
re dsoa orgosadternering o services. i noe
A basic criterion for the determination ot the period in
whtch income may be reqarded as earned may be stated as toi ows:A
Icmshou.14 aot be me am ed as earned until an asset increment
has Sen realized, or until its realization is reasonably assured.
Figure 3-25Extended TELETEX -CR/LF Option
Document A
3-38
%
12:S3 P14 FRI.. 21 OCT.. 1983PLOT <GRAPH > STARTING AT PE. I (APPROX.) RECORD LENGTH 16U0
. ....... ....... ....... .. ......... ......... . ..... ... .....
I~sCOSTS,
1s4. ..... ... ........ ....... .. 9..1 19. 197 ... 7. ... S ........ ....... ...................
.............~ O U T PU...T........... ... ....... ...... . . ....................I...
-4- . . . . . .. . . . . . . . . . . . . . . .;.. . . . . . . . . . . .. . . . . .. . . . . . . . .. . . . . . . . . .. . . . . . .
So o. .. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . .... . . . . .. . . . . . .. . . . . . . . . . .... . . .
46 0a . . . . . . . . . . . . . . . . . . . . . . . . . . - .. .. . . . . . . . . . . .. . . . . . . . . . . . . .. . .
30 4 .. . . ... . . . ... . .. . . . .
LINES PPROFI71S
2006"~~ ~Fi ur 3-26 ..... ........ ....... ........
-~ ~ ~ ~ ~ RP Graphic -- --1016?~Doumn A98 16 99 1,1 1,2 17 94 17 96 17
3-3re9 -2
.7-.
FIGUE 2 XTEIOR IEW PROPOSED BUILDINGI
Raei are runin ou os a i andnauraags. es.rit
2!Elj ears o r moareser kes over ltte cosferencenin th ng
-. trans ortin li wil grow 5harpl as demand builds. (Mucn or the]coal fCiclut to reach,too.)
Shoul a crancnetits 00ilers to coal-burning trooil or natura as ring in man cases the answer is yes.1
TAN
5'cOel~naf 1
85*
Su - ll9 oLAqJ.n
aceVN~~vi
Extended TEDLEE RLFOtoOcuen
'4TC
-a 3-40
* S a - . 5 ..-.-
.. 0-
. - . . .- - .-
L E
Bulig rpi
3-4
-. ' I12:54 PM FRI., • 2 OCT.,* :58:4. PLOT (BUILD '. STARTING AT "EL. * I (A*PPROK.) - RECORD LENGTH IEile
'I
t,.
-0
't'LINES READ - 714.
~Figure 3-28!I Building Graphic
Document B
.9
4.
' . . **. • 4o •* ° ' .o .4.. .* ........ . . .-...- . .. . - , . •..
''" " """, ,-- , . " " " "" ' ,"""". "4- i". , ."*- " .. ' .... ,. , , ' ,,.,.,' " ,
a-' -7 11 I I . .II
' a.
121S4 P" FRI. 21 OCT., 1983SPLOT 'PIPED STARTING AT PEL * 1 (APP*OX.I - RECORD LENGTH 168
'I.'
L1'COL RrAD - 714.
Si.c
3-42e -2
Piin Gapi
A. ~ ~ ~ ~ ~ ~ ~ ~ ~ D c m n B~~*. .. ~ * * 'j* . . ..i. 5v..**'a- * **. .a
3-42* * *- .
C~rrq~e_- s~-,r~~n1_3w oia e but It -must-be recoqn izedI
~~pat7s thJea -o fassets which are ntp ipcme Obviouslyj
Cant _Ino rom -a - is notincom t
i~ioa7pn ~ s s iegrJas ncome an info po p
-sits 7hicbh-iso set-Syariicrease inliabilities._ Incame
-n ---
- - - - -
P W-othhijrtry Trom customers -ZcTc~iintT, -anT f- 761att6o1
L96000
li lf N ... ... ... ..... ... . . ..................... .. ............. S L S . ....
.. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . .. . . . . . . . . . . . .
990 ..... . . . ...................... ' O.S.
s a ". .. . . .. . . . . . . . . .. . .. . . . .
A-. .. . . . .. . . ... . .
PROFITS
- - - - - - - - - - - - -
£96? 1%8 t )611 iL£'S Is(I 1972 9 192 .J7?4 lvn [~ 197 ? 97
f'PAPHIC5' OUTPUT
lthe~spal 7 f : od ; an'd _ehe enriSosrvces. If incomej
term FrTf iT snotpr-o~er-t inom de;rived frM thel
Wen-eri-1s~vfcei-
LA-a-i'..C7i-rionr fr Th ;'d eter inaRt o of -the EerIod i
ihCin'come ma~be r~earded ai earnedrnm b -e -ttdas -f01-follw:l
* In-ome shouldrnot e-re.iae4ii9 ;ai-e~r~t-1i an-aset increien~t
Fhas-ben r7alizedr-orut-iT Ttrr-ar~z~ti 71j is -esnab1 asurdl
Figure 3-30Symbol ReVp'~alHybr id
Document A
3-43
4FFF
4.. - - - * - -
-- -- - ---
WvctY m rs orl mo ree m i2-nr w Si eevair~ -In-ail ufm Wy ep ~ -Tmmmum.w
06 A w e w migni to p IE - 1e .np M 6r, ee-- w rI e-enspimep me em pmtp
IT Mei-t~~~~~ T's coal - i orWrati WW-; etiae hat we av
TE e re unn hu t~ i-adenural a. W(he fthe tl
Irun. As we begin joFeLrmor too.z~ ad-orac eere
IS ro urd - convert its Fo'iTej-s Toc~~-, cT Ea-td .re~L -t2o2' uE Lgjloj~i o. ntuaj.j.-bUrnn Dr~ran~casgs theranswer-is yes.I .
T T HaOT ~LD AIR1
SOPPILeS ag': helin4r 69% C09,11ng deina l
COLD RCERevOtp
C.OOL TOW R
IT Op
*~xr WUE P ATER A,
Figure 3-31Symbol Removal/Hybr id
Document B
3-44
A1. r-4 r, (%4 qw u-I N% %
cc M' C4'.
41 lw N-141 -q
(.)1
c0r- 0 u-i
to '0 %0 m o -I 0
00 r_%DG
00
rn r- '.0 rn0
"4 C4 >.
ra 0
4-V 00
oi 0 .0 0- '40
41
Co >
oo 41-I 44 cc
. 0 *-- W"4 . 4 0*-4A0 u-0 V:rIc
to a 41 m x *.-
'4 0 to-4
4 1 00 -~
0 (a 4 O
41 41 - .4*e4 4) 4, r-
a, 0 0
C 0 4 VI
0 0) 0 0a4) 0 .~ . CI0
4, w a ~ u
0 >13-45
LI~~ ~ -00CoI
r-4 C14 0
&A4 r4 94 r4 P1 C
4%1
0 41 -4 (
-.4 r-4
4V1
Go 0C. co 'A El
4P4.1 (4 r-4 r-4 4
F.4* A r .ff 4
.4.. 410
4.)
"4 IQ%-4
to 0 -4 v.4 (a-.
0. 0
41 '0
0 45 o4 41 0Eto 0 41 4 u >4
.. 4 >1 0 ., -0 ED
.0 r4 .041 a pq C r41 W
oo 0 >1 r41 0 O
4. in 41 414 ra .. 4 C441 0 41 to041 41) 41 0 -I
Va .4 4 9-1"4 C A., U 040 .9.0 0 4 -1
guI.4 -,i 4
w r-4 f-4 41 0'l "'4 c
45C 0 414
W) 41 0 r. 00
A- c -
0 P-15' - 4 C 41 0 41
+- 41
0 t4 0 a
3-46
'4.
(A L1&0o 0 FI 0A D (C4 c" 00 t ' ID
Ca ~ ~ ~ L %0I 0 I . ~
S4.1 C4 l (Y 0
r4 0 ~
~J 0Eu-4
01 0a1- w.1-5
4- 11 C41 - V' %D Nl qw In D3-M m.w M -4 Nl In c4 In
09' 0>i
01 rj-A %a C4 N o
Li -
.0 0 CL -
50
to V
C4C
ccI 0 I.8a.0 -4
00 .0u ~ 04 0
go A- cc 1
0 A 0 - i 041
0- (D u SVo+ to W440
+4 41 V w
0 N. - CD
>10 0i
3-47
In
0o Ch r C4 Ln P-1 r-4 C4 C4'4*1 r.I M r~I- u-I ~ ' %
4'~~L %DC or
U0>14.4
04'4
4' C4 fn at v-4 u-I a% rr ~ - C4 a'-
Ch C
0 C40
*P4
E-4
'-4~- L)- coo ~ lI
0 c0tr-I 0
>1 r-I .0
5.4 Q.9 Q . 0 10
C1 0
A* M
cc 00
*940 .0 * 00 0 N4 "4
0a- P4 $40 00 In 0
.E-
to 4' aC to
0 Q 0to-A O 0- 0 w *.46
0 C0 01 00'-~ 0 r'ui
0 348
lapha 0
4.0 Task 3 - Analysis of Results
*4.1 Compression
The compression estimates, calculated in Section 3 for all
- four segmentation techniques, are summarized in Table 4-1. For
comparison, the compression for normal Group 4 Modified READ
(k-C, no EOL) is also included in Table 4-1.
As expected, the results support the general conclusions that
with increasing graphics relative to text content: the amount of
compression decreases, the distinction between the four mixed-modeS.techniques diminishes as does their advantage over straight
forward application of Modified READ (k=W, no EOL).
While the Symbol Removal/Line of Symbols technique provides
the greatest compression of all, it is not robust with respect to
arbitrary location of symbols. The Symbol Removal/Scan Line does
handle the case of arbitrary symbol location but does not take
advantage of the occurance of lines of symbols. Furthermore the
Symbol Removal/Scan Line technique provides the poorest
compression performance. The Symbol Removal/Hybrid technique
remains robust with respect to arbitrary symbol location and this
technique also takes advantage of the occurrence of symbols in
lines thereby providing compression close to that of the Symbol
Removal/Line of Symbols technique.
The Extended TELETEX with CR/LF option provides compression
intermediate between the Line of Symbols and Hybrid Symbol Removal
4-1
A,
Go 0 0 0% VA- -4 0 0 o
r-
0
ra .0cl Oa4 r- 0% %D LA %
0 -o
u lz C-4 C4 r'-4
004141 0~
2 r-. j 0% LA Alen c0 "r '-4 n -
- ~41 0'
51 ue- 1- a 4 en 4 Ln
"I
-II LA%0r:
C0 r- o o Ca A C4S0 0 LA t-1
0 z"4 -o
E--4
0~ 0
4.4 CO -I0 ug0>90
E-4->0
0.
0 o
>1 :r) U) Cz
C4-0-6 40..
#- IL %
techniques, being marginally better than Symbol Removal/Hybrid for
the more Graphics intensive computer generated documents and
slightly inferior to Symbol Removal/Hybrid for the CCITT No. 1
document.
4.2 Complexity of Implementation
Not much separates the various techniques in complexity. In
all cases accommodation of scanned documents requires symbol
recognition. The three Symbol Removal techniques have the same
image storage requirements independently of whether the recognized
symbols are to be organized into lnes of symbols or not.
Generally the 32 x 1728 = 55,296 bits required to accommodate the
larger fonts and hang-down characters will permit symbol
recognition, removal and organization into lines.
In the Symbol Removal/Scan Line technique each recognized
symbol is incorporated into the transmission as the recognition
occurs. In the Symbol Removal/Line of Symbols and the Symbol
Removal/Hybrid techniques provision must be made for calculating
the linear regression of horizontal and vertical symbol positions
to identify line skew and perform the vertical realignment, of the
recognized symbols to a baseline, necessary to remove the skew.
Extended TELETEX, with CR/LF option, is presented as a
technique for generating a mixed-mode message by a computer; not
as a method for scanning a document and segmenting it into.g
ct
" graphics and text.
.. ; 4- 3
9'-1
4.3 Commonality
This section discusses the commonality or ability of a Mixed
Mode machine to transmit messages to, or receive messages from
- - such existing machines as:
(1) TELETEX
(2) Standard Group 4 FACSIMILE, without
mixed-mode capabilities
(3) Group 3 FACSIMILE
Changes to these machines are not considered permissible because
they are already in the field; rather, here are considered changes
to Group 4 mixed-mode machines necessary to provide commonality
with existing machines.
The basic core of commonality between the existing machines
and mixed-mode machines is built-in in that all mixed-mcde
techniques considered use the TELETEX code and the Modified READ
II code proposed for Group 4 FACSIMILE machines. The Group 4 code
modifies the Group 3 code by:
(1) Using k= instead of k=4 for 7.7 lines/mm,
(2) Deleting the EOL code for each line,
(3) Eliminating bit stuffing to achieve minimum line time.
General compatibility betwen Group 4 and Group 3 machines,
with respect to the encoding differences, is expected to be the
rule rather than the exception. Therefore commonality with Group
3 machines is implicit in the discussion of commonality between
standard and mixed-mode option Group 4 machines. No special
4-4
I.VA
a-
discussion of commonality with Group 3 machines is necessary.
For all of the mixed-mode techniques to achieve commonality
it may be necessary to inhibit information about the stored
library to be used. Other than this general requirement, the
*• Symbol Removal/Line of Symbols technique requires merely the
- inhibiting of symbol recognitions to produce Group 4 transmissions
.. while reception of Group 4 transmissions requires no modification
whatsoever.
In addition to the above, for Symbol Removal/Scan Line and
Symbol Removal/Hybrid, code bits that change mode must be delected
on transmission and inserted on reception. For both techniques
this is a single bit that precedes each scan line.
For Extended TELETEX, a Group 4 transmission can be obtained
by inhibiting all symbol recognitions, including blanks, which
will force the entire line to be transmitted as graphics. In
addition, the codes for the last symbol on the line and the
graphics width would have to be deleted. For reception, the last
symbol code and a graphics width code of 1728 would have to be
added before each scan line to correct the Group 4 transmission
for compatibility with the mixed-mode receiver.
For Extended TELETEX, to receive TELETEX transmissions no
change is required except adding the code that identifies the
stored library to use. In transmission, the graphics mode must be
inhibited, with space symbol codes being transmitted whenever
material that cannot be recognized as symbols is encountered.
Also CR/LF codes must be inserted for each line.
4 ---4-5
For all techniques except extended TELETEX, in transmitting,
the graphics mode must be inhibited, and a blank symbol used to
replace each 20 pels of all-white or graphics pels. Also CR/LF
codes must be inserted at the end of each line (approximately 33
scan lines). Corresponding changes must be made for reception,
namely adding coding for approximately 33 all-white scan lines for
each LF, and deleting the CR/LF codes.
In addition for Symbol Removal/Line of Symbols, the 12-bit
(EOL) code that indicates a change from graphics to symbol mode
- must be deleted on transmission, and added on reception.
%4-
:::4-6
"2.
* 5.0 Conclusions And Recommendations
Figure 5-1 summarizes the subjective evaluations given to
" each mixed mode technique for each of the topics considered in the
study. Note that there is a slight preference for the Symbol
Removal/Line of Symbols technique relative to the other three
algorithms.
This study assumed that the OCR function was performed
perfectly for the scanned input document. AT&T has submitted a
recommendation to the CCITT for a mixed mode system describing a
specific approach to the OCR function. It is recommended that
" . this algorithm be simulated in order to properly evaluate the
impact of a realistic OCR system on mixed mode performance.
,5-.1
4->.
-'.4
• . , . - . . . , , . , . , . , . - , . . . . , . , . % % % , ,• , ,, -
0 .4.0
ci 0x 0
0 >
r IIS 4) 40
0)415
040
0 0Wl a)50E-1~ r1.4 0
150 0 > E-
I->C 0
0 0
0 0 -I 0>5 0rx. 0m 0 E- C.
4.V .1. ?a ~ Q)
r-4 -- to 0or
C - 00
C5-2
a.
-m
.:A.PNDIX
"-.% CCITT DRAFT RECOMMENDATION S.a-W- RTOO
' ~DOCUMENT INTERCHANGE POOO
-'q FOR THE TELEMATIC SERVICES.1,'
S.'
I %'. '
i a - _ . . . . . . . . . ., . . . . . . .. . . . . .. .
.o.,.CCITT
CITTemporary Document No.
STUDY GROUP VIIIWORKING PARTIES VIII/2, 3 AND 4
Geneva, 214 MAy- 3 June 1983
. Question :
SOURCE : SECOND DRAFT OF DRAFT RECOMIENDATION S.a
TITLE DRAFTING GROUP ON DRAFT RECOMMENDATION S.a
U* -1
lo- ULLe -TttAk..S+ C., V%, S
,
. O * t , p - -,.
"2-'
frd-j for&
-. 11
1 4.1 c 'eA- 4----
4.3, s
ZJZ 83060Of-1
a t4 .C
,
0,1 W~z10-
J*Ntv Q A tnO.S%,.L
A- 2
Ir....t :I.., -~. P.: -er'" for the TelemaLLC Services.
1) General
9, 1.1. Scope
1.1.1. Concerning the service aspects
- Recommendation F.200 lays down the operation provisionfor the automatic internatLonal Teletex service. Theservice requirements unique to the mixed mode ofoperation are described in the annex C of the recommen-dation F.200.
- ecommendation F... should describe the service requi-rement for G4 fac simile service.
1.1.2. On the technical side
1.121. The Lerminal equipement is defined by
- Recommendation S.60 for the teletex,- RecommendaLlon T.a for the GA facsimile
1.122. Concerning the information coding
-Recommendation S.61 defines the characterrepertoire and coded character set for theinternational teletex service,
-Kecommendation T.b. defines the coding schemeused in G4 facsimile equipments,
- Recommendation S.0 defines the coding schemeused in videotex services.
1.123. RecommendRtlan S.62 specifies the control proce-dure for the teletex and G4 facsimile services.
Note : the generalized session protocol, underdiscussion between CCITT and ISO, should be
also coasidered.
1.124. Recommendation S.70 specifies the network indepen-
dant basic transport for teletex and G4 facsimilei services.
.".i This Recommendation $.a defines the
P ot reL to be used within the Telematic Services when a docu-
ment structure is required e.g.for Mixed Mode Teletex and
for Group 4 Facsimile. (w~ 7 A'C CT )
S.a is embedded in a Framework of Recommendations for Tele-
matic Services as s/ow? /n T ,p'c
A-3
,- ",", i( f , ". .',." **% ' ." " ,"9( .',." " " . /" " - . .M; . " -. . . . ." • . . - • . . ,
z • - . . ' . '. . -q; , . -. - . -. -
L7 77
-4-
" ' S.60 T.a
S.61 , T.b
Trx',\ Mode
.S.a
S.62A
L S.70
"u. 1o 'Framework of Recommendat tonsfor Telematic Terminals
4A
i A- 4
"."". For the ,ose of this Recommendation, a document
is ahz.! :oeur'text that is interchanged betweentelematic terminals.
/..2'4 A document can be interchanged for two major purposes.- It may be interchanged as an original in a final form allow-_form allo.-
ing for printing, displaying and fi+,,g at the recipient.- It may be interchanged in a revisable
form allowing For prOce-.si,1 at the recipient.
Processing includes editing, reformatting, filing andother manipulations.
-. 2,1. Text is information for human comprehension that can bepresented in a two-dimensional form, e.g. printed on paper
or displayed on a screen.
Text consists of graphic elements such as character boxelements, geometric elements and photographic elements,
*. which constitute the content of a document.
i ~co,,. ..-i.
The.Am of a document need oe separated into( varions portions in order to :
- delimit *1w presentation ---.. '& _ j 7 c /Vc- ,4d,; I03iverl e%/'C/s . C'C( , pvajr,,sT
-use different types of coding- allow processing after commaunicat ion.
74C e CesErp' 1e g( /4ese P/"&_S if/e eb' 7
.., - A-5N44'.. ' .: : , , : , ; % ) v < . : : : : < - . - . - ,- . - . . . - . . . . . . . . . . . .
The layout structure relates the content .La..4 ocuinant cz ru CL
fey/ be ?poitioning and rendition on the presentationmedia.
The logical structure relates the content e ~ianto 1oqical text objects us 'rinservingspeii prposes: sections, headings, paragraphs,footnotes and figures.
The architecture that is particular to a given document is
called specific document architecture.
Vrrsi~ 1 ,/js foS ~ -,4c - 7~ h alot~pear re'vpaf
7L47e-S /nj 7/A/,_ 'Cf7'
The
/4I jeP7_eWtC 4 5 O)
a ~ ~ ~ ~ ~~~~n .'a m,~e~ f'r C '1AY( 'fyc#/ bfrj-/
The interchanged generic document structures help
-to improve the transmission efficiency,
-to maintain the consistency of the document with the
document class during revision at the recipient and
* - to facilitate the creation of new documents of a certain
class.
f'reJoyc Q e(r,'' de-s'P'O&i oF 0 dcubklcni ' ~~
oo P~ rs( r srcA- a#,djep r 'cvj ?S JAW
/4W4Oi e cl t7le r.S e1Me C/s /. P?. /,,a / e 7 7fl ez 1'Y7
-'li
Tw e *, /I WO, ,/ r
- */-~' Jon,~/ A-6
AM .1*'.
. ,7
.-..-.
document
uuenc voii specific:
-.,-. -..,
.5,..5
loial - - WN ogcl - o lyu
-- sta*drd paragrphs, looncarce bo-€c, geometrc elemns,,hotograhic lemennts
.- ; )gure 2. Docurnent Archttectu re
ZN
.A-a,
SA- 7
V , ' , , ., . . . . . . . , , . . . . . . .. . ... , -.. .. .. .. , -. , ..
,5I,. , , , - ,,, . . . . . . . . .
1.2.2. Communication concepts.
1.221. The capabilities of the rr aL~.p protocolN ~ re- are negociated
In the session establishment phase.
The terminal capabilir~ies required In order toreceive the document are globally indic~ted by the
inii-. rtcl parameter.
1.222. The necessary elements describing the 4 0cument/4tructurejare defined in this Recommendation S.a.and will be carried inside the the Sesion ServiceData Units (CSUI-CDUI S.62 commands).
.A-8
V:.
, 1.o2. CmmuntaA-Bo
LALCA LAL CQ-
(,kz f, _ _ _ _ _ k^u 4t1
7Lf. a L& At K W -6-e v
A-9 8?3060 1 -z
~~~~~~ Yjsbo e de3-
A OL-c e~'-~ .
Z .Q-, s iL r £Ll, rj (Ur O zc.
s~w) G) v-eA"-bt
ke . ?d
-eA- al--
0£r-e o LL el C3 a t 4c s.
It cc~A of-, 4-)
IPL60TCI l 0-... C
A ILcL v L s1LA.- - cL cL s 6ra sp c-r e-
Cr (~L,4L~/. lTL-IJ c.cz(i*~ c&-
* y -~4( tA, 4yw d"> ~c 1 .
4j'2 63o 6o iA-11
A (
~Joec
0+*, ( IvE t e' See ~ c.; (y.)
- ~ ckr-cr 5{4fec r&) (O4 OcLa. r-
St LcL,,,a .. 40 L4A r7 4 u
6.4 P".14 s. X
&pi e-eec ck e A-LL L
* oj a.-
e%~L C3 -s-&p t * cjG (
U.,~ A-- LL.8oc ICS
* v4'6er648 Oflrltkes .0C4A~- Le
sf ealji4.0 ,el.D- e** '' for 64 e- 66j e c
o cL L rL~ -c d.cf/ ~A4 Lgtu-.r-
IXe s eOLJAf a-e...u
lay~~~ Z e,- sAc4o. c~4Zk&,~*~~L
Us LLoJh -z4- kc p-,jov&e rtj0. f
A-13 Z 6'3IO oi- f
e2
t~r- c' 4---mU-
-4-c
JA-
A-1 Z 2 (900
J 7~~ .i~* -. r .. ~~-r.N7 -V -.- j-.-*..7
LAA- C ",CI--
21.rt-3±t AE*bL r(~O%
OCOL&- "J ClCL c",, A
LJL44tAlI, d C '. )
6l.3ecf.ej C. Lb IA., r -L CJ6- ;e -e
4~T4
A-5a.o CD
5 r 1m-c*E Le C/cs
~~Le.LP CAL-a~ LZt AeS
T-c C.1 j c
eta.. a. 5~ed c L4&
.9'--.'-
ISOeA i,; ( A. -JL&'JtA- 6 .. o.4
-7-
4.
*T A* LAI 0ro LOY
~C k
ouJLet~y c -- , C.
TL-)A. kA-./
*e r k. tL4 -4)er.-
A-17 Z)2 c?%o~c,
-~- 0-0 COI4 .,4
A- 18 )2 8.3o60
-t -~s -c EA.
' '
-C'V Zs Lk A-hI-
4..4 C413 E-,L cc 6 L
A11
* * - . - -7777-7=7 7.
a'..-
-a..s
" 7a
1Of
Apcr- IU (tP 4- . -A
* 6o &s4oz A4*~t r'4c & 4, ~ 4-e-,~
2 ~ Tv A-~.~-~- (-~t
"o I k- O4-4 St c1.14-4 W,
cL A ~ 4bv, cLv-ir-&t(,&.r 4& OF
-~A-20 LJz 83,060of- 7
2k ~~ rpJr~ &~'~ ~L'..e
A ~ ~ ~ ~ e,*4 or~,c oir -- ~~_
A. ~ la,
As~~~1 2 4j.L L,:, &,j 4-t 6Ir
A-21 /32 830c) 1-8
b) C. 3 rre~e34 ~E~~L IL dA. v~C djc~
)rot,~. -L .- J...
c)~~ ~~ Co-fo-r. 'p- A-- 4l~c C OLt&4S
* ~e4- L4~~r~ 4~r-~p0~ ~T'-~euo
s d~ T"L t I - i, .'e--U-
4L,
.-...-. 3 0'I :.. 41-
-. ~ ~~~~ 7-v'* w.---7--- 7r~ ~
,-Ncf, Aut.Z6e -AvI t' Se-.A2-
dA& W
A- 23
D-8037 592 MIXED MODE FOR GROUP 4 FACSIMILE SYSTEMS(U) DELTR 212INFORMATION SYSTEMS INC JENKINTOWN PHR ADEUTERMANN07 NOV 83 NCS-TIB-83-7 DCAi00-82-C-0847
UNCLASSIFIED F/G 14/5 NL
EEEEEEEEEmnsniIEEEEEEEEEEllillnIlEEEEL
L4.0
111L51.
MICROCOPY RESOLUTION TEST CHARTNAWWF4A BUREAU OF STAN4DARDS- 1963-A
*NI.
*.*.', 4 4.r
.f~~~~~~~ %** ~ *9~ ** '*
%v IV%~ . *% % . ~ '.~
o#,w-A4+r i,M 6
AA C
bL.CA (3)LI greve e(0), apJc (4) 1
Al es4oy. * . &4
-Jo$*-rA~r;6k s *.-
PW 6a b4
042rL&44r,6kC f*4,%.0 ~c,
A-2
9,:
opec-wc *l-r ar xvzt- r r r.
cmosj ed. #
&el). Ile 4ttvm Joer
1'k 4 vk
(crv 2A ) 4aA4pCL. 06 A A4e-c 4- ~
Mf~Q&ef4C 04V ,
-t 4 o~&dL
A- 25
Ve otN6 .1 a pav or .A are
Ad o4"j* fv-t o6ec*-
A; lbAt A# 104 Ift
-totA4 "go L46tg
A-2
-2 ~
% a Jm*C t k;26 ape ^16.
cLraI~P'4t A+ci
IC.j~r4 - - 0 .. 27
M 0,eISO a.9.
-Vsr
~~'~ -- 4
A- eJVr 27&
* ****~:**.* . *. ~ *~*.*~A-27** *
_*--* - --
-
-i'- V I- V.
Annex :~A~. 4,
* Iocumenv ayh&vt04iecar ra
_:4.:
. Document architecture
1.1 Specific document architecture
1.1.1 The architecture that is particular to a given document iscalled slncific document architecture. It consists of the
following components (see Figure ,):
- content- layout structure s specific
- logical structure document structure
- layout directives
.1.1.2 For the presentation on paper or screen the content of a
document is physically structured into pages, blocks, lines,
character boxes etc.. This structure is called layout struc-
tur.o and the objects building up this structure are called
A-28
I~~C Aunmbu ieipfian of adocummo
[rminmbawom upudn *k dwoM nuhn
*N $bd Lnai L i~f g LcEiIpmV IJV
-*4 -J F bi" ama~- h
LP~N -bwAaona~ W
Pabl - _ _ wf
riLgur. Document AwchItecturs Model astz-uctvre, runctlonu
A- 29
6~ - 4 4 4 k * ,,. < . ,L .. ,, ',. . . -... ' * . . *, , .*.. . . .. . , . . - . . . . ., . . , .
.qo -
layout objects . In the final form theEl content is divided into portions which belong to this layoutr ,¢.',-objects.
.4.1.3 On basis of the final form only such processing can be done
efficiently which causes no reformatting in the environment
of the manipulated layout object. It may comprise layout
revisions like to scale and to move blocks within empty
space or to overlay them transparently or opaquely. In cases
where the environment needs to be reformatted, this has to
be done manually by the user.
.1.1.4 Documents use to be logically structured in order to enhance
the comprehension of the text. In the final form of documents
the logical structure may be implicitly expressed by the lay-
out of the content, i.e. its arrangement within pages and
its type style.
.4.1.S In the revisable form the logical structure of a document
is explicitly represented by loqical objects, e.g. like sec-
tions, paragraphs, footnotes, figures etc.. The content is
divided into portions, which belong to the logical objects.
The logical structure can be edited (revised) by commands
like Oinsert, delete, move" etc. applied to logical objects
like a paragraph. LiSure 'I shows an example of a specific
logical structure.
.1.2.6 In the revisable form layout directives can be associated
as attributes to the logical objects. These layout directives
allow for the control of an automatic formatting and layout
of the content portions belonging to the logical objects
during editing (revision). Such layout directives may be
'centered, left aligned, two columned' etc. and "emboldened,
underlined, italic" etc. applied to paragraphs, sections etc..
A-30
L! l.ir ! lil l-l -~t' g ,<d,<,'..,-l"' .',,."' .'%,.. '.. ".* r '. € ,,i.'-., " /,..'. ". , * . "( .. ,.".,". ",. l',-,. " "- ".J"-"."
4
* 0
*0
0
OE3a
Om..4
A-31'
In a given document common content portions may occur se-vera. times like a log on several pages or a -standard pa-
ragraph within several sections. For purpose of transmis-sion efficieny the common content portions need to be trans-
mitted only once in that part of the document interchangeformat, which contains the generic document structure. Inthe specific document structure there need to be only re-ferences to that common content portions.
4.2.4 The generic layout structure defines layout templates for
pages containing the position of predefined blocks with com-
mon content (e.g. logos) and of hframes (e.g.the image area,
an address area) within which the content of the logical
objects may be dynamically formatted. Such a template mightrepresent a standardized form like ISO 3535. Figure y shows
the example of a template of a form.
The generic layout structure allows also for predefined se-
quences of pages with predefined layout, e.g. a template forthe "cover page* followed by a template for the *introduction'
page and a template for all "nw section pages" of a certain
document class
1.2.5 The generic logical structure allows for the definition of
types of logical objects, named generic logical objects,
which are characteristic for the document class, and of the
definition of their hierarchical order and their possible
sequential order. The hierarchical order is expressed by
a consist of' relation, e.g. for a section consisting of
none, or one or more subsections, which may consist of one
or nore paragraphs. The sequential order is expressed by
a "followed byO relation, e.g. a head followed by none or
one abstract, followed by one or more sections, followed
by none or one reference list. The generic logical struc-
ture can be regarded as a set of rules (i.e. a grammar) from
which specific logical structures can be derived.
Fiaure - Shows an example of a generic logical structure.
A-32
,.7 . . . , : . ., ., . . . .. . .. .
N The formatting-and layout process creates or modifies the
layout structure and results in layout objects like blocks
arranged and styled according to the layout directives.
.1.2 Generic document architecture
.4.2.1 A given document can be regarded as a member of a document
class like a bussiness letter, report, purchase order etc.
The generic document architecture provides the user with
means to define rules for the logical structure and templates
for the layout which are characteristic for a given document
class. A document class is defined by the application. It
is not intended to standardize any document class by Recon-
mendation S.a.
4.2.2 The specific architecture of a given document can be built
according to the rules and templates of its document class.
They are described by the generic document architecture of
the document class which consists of the following components
(see Ficuret.)z
- co mon content portions
- generic layout structure generic
- generic logical structure J document structure
- generic layout diretcives.
( The interchanged generic document structures help
- to improve the transmission efficiency,
- to maintain the consistency of the document with the
document class during revision at the recipient and
- to facilitate the creation of new documents of a certain
class.
J.2.3 The comon ,antent portions of a document class are prede-
firod port, is of text like the geometric elements of logos
in .. s, ahe character box elements of standard paragraphs
in a6i..ority documents etc. which are comon for all specific
documents of that class.
A-33
- afic
.q% ~ -- 's i ~ ~ ~ *P.~ V.~~~V~% .- *
.1*.Zi.,7
'.48
PROVEN
I 11
a tem lat o a l t r r
I34I- -- .. .t~'s I --' ,' I -- - - -
I - bfD - .., d ef ,q d 4V/u r I
a'. I
* I
* I
v- I -'a I
~~~~PROVEN ,,A.... .
rP:guze 3. Example oL a teml3ate of a :ltter feo
A-34
a "+ +" ' , ," "--"4~Saa*a . '. "'4',:- " ' -' i:;
4 WN W.I W '
" . , - 3 "
...
Eq 4
.,'
..S.'
A-3
.4-.. , ;\ . -. @-; , . -... , ... . . - .
, - W. 7
S.2.6 The generic layout directives are attributes of the generic
logical objects and apply to all specific logical objects
of the same type. Similar to the specific layout directi-
ves, which are associated only to single logical objects
the generic layout directives allow for the control of an
automatic layout of the content of logical objects on thepresentation media. There are twqo types of layout direc-
tives, the one effecting the positioning and the other ef-
fecting the type style of logical objects. Specific layout
directives may overwrite generic-layout directives. If a
given document has been changed by this way it is no more
a member of that class from which it originally has been
derived. Fiaure rdemonstrates the functions of the generic
structures.
" ...
~A-36
V. %
d-,i
l." .4
~A-36
- .s .-
* . - . . . . . . . .
'V specific logical structure specific layout structure
--
"oR
A=-- - - -I
A-4.3
0
- 1-4
gi aION
S I
genri loia trcuegnei aot tutr
a- A-37
o',
i-Ii
interchange formats
.2.1 Categories of interchange formats
By apropriate selection among the components of the docu-
., ment architecture different interchange formats with dif-
ferent capabilities can be derived.
Two major categories of interchange formats are distingui-
shed
- text imaging formats (TIP)
- text processing formats (TPF)
.2 Text imaging formats
.2.1 Text Imaging Formats mainly support the imaging (printing,
displaying) of documents at the recipient. The document is
interchanged a an original in the final form by interchan-
ging its content and its layout structure. The content and
layout structure of such received documents can be edited
(revised), however only by doing any reformatting manually.
The test imaging format offers no support for an automatic
reformatting.
There are two text imaging formats. TIF.1 (Basic TIP) and
TIP.2 (see Figure ).
.. 2.2 The text imaging format TIF. is called Basic TIP. It con-
tains the content structured by the objects Opages", "framewand OblocksO of the specific layout structure
The Basic TI? represents the final form of a formatted
document and allows for an exact reproduction of its image
at the recipient as created by the originator. There is atleast one frame, which represents the image area (1 printable
area).
A-38
* .. - .~ --..-. ~ A- A**.*** ** * - * **~*
A..
-- A
'-4
-A
=* 4 K - -.#1
* - *k~ m
-'a
* I II ~fI#i#=111111 jj [Lg:g.+ - - ~=
-j
hiIi'
4
4
i Jil I9. ill xx x
Iiii I5-
4 U.III xx x x'a..
0
f~jx x X
* iii ii S.1~be
a
A- 39
* ~S *W -~ 1 -.- 'q * 45 p.5~ . ~ ~ "4%%.% ~ *..
* S A q .* *
-4-
For the easy repositioning of logically related blocks which
need to retain their relative positions during layout revi-
sion at the recipient, there might be additional frames which
enclose these blocks. Such a frame could be one enclosing
a diagram block and several caption blocks of a figure. How-
ever these frames need not necessarily be interchanged.
Note: In the context of TPI frames have an additional essen-
tial function. They define boundaries within which
the content of the objects of the logical structure
can be automatically formatted. Therefore layout di-
rectives which effect the positioning refer to frames.
%2.3 The text iaging forpat TIF.2 adds to the objects of the
Basic TIT the objects of the generic
layout structure.
It allows for tranmission efficiency. Predefined content oflayout objects which occur repetetively on different pages,
like components of forms, are transmitted only once within
the generic part of the text imaging format. In cases wherethe document class is known by the recipient, e.g. within
a company or for standard forms such as e.g. ISO 3535, TUF.2
contains only the name of the document class within the ge-
neric part of the format and the detailed generic informa-
tion is added by the recipient.
1.3 Text Processing Formats
1.3.1 Text Processina Fornmts support the processing of documentsat the recipient. The document is interchanged in the revis-
able form by interchanging its content, the logical struc-
ture and the layout directives. The content, the logical
structure and the layout structure can be edited at the re-
cipient supported by automatic reformatting.
A-40
There are three text processing formats, TPF.1 (Basic TPF),
TPF.2 and TPF.3 (see Figure )
3.2 The text processing format TPF.1 is called Basic TPF. It
contains the content structured by the objects of the speci-
fic logical structure, the specific layout directives and
the objects of the generic layout structure. In the case
of TPF.1 the positioning layout directives refer to the
generic frames of the eneric layout structure (e.g. the image
area).
* The TPF.l allows for transmission efficiency in saving the
interchange of a not yet final layout. The unformatted do-
cument can be formatted at the recipient according to the
specific layout directives given by the originator which
may be accomplished and revised by the recipient. The con-
tent, the logical structure and the layout struture of the
document can be edited together with an automatic formatting
according to the layout directives.
L.3.3 The text processing format TPF.2 adds to TPF.1 the rules
and objects of the generic logical structure and the genericlayout directives.
The TPF.2 allows for the interchange of unformatted docu-
ments which can be formatted at the recipient according to(the generic layout directives given by the document class
and the specific layout directives explicitly defined by
the originator. The document can be revised under control
by the generic logical structure which helps to maintainthe consistency with the properties of the document class.
The types of logical objects which may occur in the speci-fic logical structure and their possible hierarchical and
sequential orders are defined. In order to format the docu-
ment, layout directives have to be added to the logical ob-
A-41
jects by the recipient. In cases where the document class
is known at the recipient the generic part of the format
contains only the name of the document class and the detailed
.generic information is added by the recipient.
It.3.4 The text processing format TPF.3 adds to the objects and
layout directives of TPF.2 the objects of the specific
layout structure.
AThe TPF.3 allows for the interchange of an already format-
ted but still fully revisable document. it adds to the pro-
cessing capabilities offered by the TPF.2 the capability
of reproducing the iuage at the recipient exactly identical
to the one originated by the sender. This format may be named
*full text foruat= .
A-42
... r ' -p - - - - .." .9--.. - .". Z ' * ',' '_% 4*%* ",. % -- ' -" ' '
APPENDIX B
Combined Symbol Matching Facsimile DataCornpression System
WILLAM X. PRATr. mn WM Smn. ]PATRICE 3. CAPIANT, ui.tnWEN-HSUNG CHEN, ERIC L. HAUMLTOtJ, un =. AmD ROBERT H. WALLs
A~-A boof dM imW Wow ioo mbbmi 2M~ me is Pk in doMy mi im bmy pua is aw-~ ON*" (I), k PUO 7f m u m eWILNbmw womb woiln.Mb~qni SOf bibmi n a no"@. ad M Goi"
.WW &Who ad W=.d aslib. In so bf =ads. a by a go~bemw , smil cedft madm. lebs W=Wa m.rNWbdf Gpw~ Whmr dowww 4 ab no~ a dpkw Wm*s asm wmW is pommn3 "i w& wy bi bb"i wI-mf dwom ud son 1nf No pw m Mw owf AMO& a&m M =W is~ ubmv eW iniU AMVI eab
NO am=" 1 ~ i a om, ni 0 no m W is b 6Mi.WNmi 6 SUO a~ fdtv SoN m -. E~ Sm NN "@MO mf ""MO f f to MEUl ib
UrN dM SO do waW but nusU eaw so mm.W go gom. is& @m& S& iiOIn ab
460b vIS Sm Of MW*00 w~@W"b a how of ftwSOPINuOu 3Cn Sup. " U. W Naw 9. 191g. OfE- *M. is SNb Wf ON dowmnf Is
W&M Wea -~f Omapms W^ be. Soso Uno Tom to-b -oo sag*. u a" a t not I ftw hamCAM64 Odin" ft 69 U3 bm =bd M
0O184219IlO#O7O4?561O." @ 1"0 IIU
L 76 J6- 75- -. P~.. ~ ~ --
PRAIT at uA: COMUN SYMBOL MATCHIG FACSIMILE 757
OSr facsimile coding systems Previously developed CDNllShave been bind on the concept of run-length coding ?VI lot.LJ 11. Ru~angsiih coding methods proide a relativelyIN Key SMBLSYMBOL
higa aop c rtio for a graphics type of document or an 4Ul SEA CH4 IOCKI M~IATION
aiphmnumA document containing a sma amoun of text(21. Did, the achievable comspression rtio. drops appreciably UNKOWN.If a document is filled deasely with alphanumeric characterbeoom= the black and white runpagths become quite short.Deos alp1 hanumeric documents =n be efficietly coded by LBAY CDN
symbol recopidons techniques in which hndividual, symbols ans umummw# CANDIDATE
detected and coded by a prototype library code (31. (41. VicCT WECOMfts y ToTEIlbwever. such -&-method cannot effectively handle doumnt UCUI "~'~. MTCHING PStTI E uPP
cotiiga mixture -of aiphanueao and paphics. One ECAINUKON1 FRMTNDF"I prsposed appioechk to thin problem has been to semeunt aPATRdocument MDt strip containing alphanumeric text or graphics ftg L. C8M comnsim.date, end then code the former by symbol matching and theletter by n4Uk~h coding (SI. The problems with thin ; PXL HUKKapproach ar the difficulty of document segmentation and the (IS. 4s_,dro in comprendon fefuec f the egmentation, a notaccuate. Thin paper introduces a new concept of hybrid I
con ding in which a document bdynamicofly segmented Into symbol and graphics reglos (61. 6
Conoepinsll. the symbo veris iiphim segmentation pro- ) .
end mployed In the fooas comprsemor is quite aimple. Adocument In earned. ea by line, end all isolated symbols that________see euFSP1ted to 9 im In the doumost ane extractedl end coded N&s 2. 00* *Ah .
by a£ sh b' matchiog process The remainder of the dciimet, called the reie in coded by twomdimenioml ruelonea coding. Thinsp segetatm nomthod , j doclmew along with the location. coordinates of the symbol. If theZ Zblet e ne by syblmathn i ' - comprion I nuosl the new symbol is zohtrehom the Paphics ordows of a documset end lIminates mitdand placed in the library. Those swen of a docuiment
insoto im. -- i n Wae t~ e bewen o9' a rd,,end a to m ae n4enth coding tech-
SThe --- bl' m pIoe, , preopoubl described lsbendeecribe lay elomnt of the oder in X1 detaiLadapted to .eo lbuel oeactu- In a documnent.kn tis symol uoPdo1n4 gade of m I- m a the documeont A. Symbol Dklmgdnin Vepe!e e by conventional wIne codes. character, specs, The fuction of the symbol blocker is to examine the inputagow retumn etc. buffier in a systematic ilcs , and to locate the poution and
The following seilon doodles the comb ined symbol match- dos of any hislted ch1rce- Fig. 2 ilustrates the blockingin (COO) dd for bath the boaile nd symbol neo- 1roc.. A block ptxel in the buffer, denoted by the character41@Wem mo"e Ot Sogmic.. MI" is consdered to be a "ey pixel whenever the fer negh-
3.F Cases. Moms bornleead aoe it nd toite left ste wte,ush c hew eowThe bleek Op of ft I deeibes the buid lmnsf of 00
doe CS softn syssom fin haie oodie Jo .oa a01nuember oeen bee wi e Bout two to lowr those theawem comoer hegt aed in a scrolled, buofter -hi Whenever a key pixel is snsomtered, the Blocher is initiated.da In then cusumimd Ie by Ife to desesini If a Mesk pixel The blushe"I Ixnc them pls froso the batter that arnes. i the each. lins oomn so Meock pinel, tis Weeoms- ontiuous with the key pixel, or encloeed by a set of cont*
Me" bn esd by en fedioe ode. If a block. ,ie su, um blc~ ias. For example, wit the lower ca letera bloeimg poesm in conducted to insdst the symbol For. "%" A~ blak pM*s end the eclosed white Islood" will bethoe heistd symbols, furhe 'rosem, in tq obe- to GXt3~ by the bIockmrdwasbe it a IeN, of the symbol uinder suaslma el-"soi Sodo in ase Nheor Thin proess isuvee as mroc. A Fourr Sxin'eramofa a atof fetre, a -6 ig operaion to rejme- The saw straightforwerd method to determine whether apeaadeo eams, end awy a seainsatf gimpisteotaboe. match exises betweenn kmown Symbol aind ose of the7Wu &a Ied dinu end It Ia. - esew e a~ay symbols Amed is the library Jo to perform a template maoch900 IM On VIMaPe NNW, and um a new bloehed dmw bteP do unknown mnd every library symbol. Howeer a
am in earning, it in eem wate a h M of the two. d epbs am*s n oodly & terms of proB-NOW dW thu do ee SMM Set N the eoir in In times end oqmn=. A maend of I Iaj the numberofswueof 60 Nhey Ideem (W) sods inu bd eac mesbse is requied. The dotnes Mess been taken in
-~~~ .. .... .~ * - ~ -
S708 PROCEEDINGS OF THE ELE VOL. 64, NO. 7. JULY 1950
I~
Aps
1k.~~ S.Fde ewr Weeraybol Wlia&, N- sbol besbt.Pu armed 00w ptwew, A a bee em pis maadd wtsaara
to extiact a set of scalar ' satum" from the various symbols A major shortcoming of the conventional template matcherIn the li1brary. These features are sed to reduce or "screen" described above is that it treats all errors alike reordless ofthe number of candidates for a template match to a tiny wheo they occur spatily. An improved matcher, to bebiaction of anl the polsibllities in the absmy described, utilime a "weighted ExcLusivE OR" error criterion
The features used in the screening p r , - are the block that is based on the context in which the error occurs.bedit, block with, symbol perimem, and pixel areaenclosd The motivation behind the weited &XCLUSIvE OR countby the noter boundary of the symbol. ftg 3 provides an error criterion. insy be appreciated by examining the EXCLU-elample of lanMWn derived fom a clumictr. 1W! ORt error (denoted A 4DB) in Figs. 4 and S. Compare the
EXCLUSIV& OR pattern for the "e' and "e" in fg 4 with theC Cmufte Se'meg patter for the pair of "c's" in ftg S. Note that the EmCu-
Vhe purpose of the - eeIng; .roIs is to veduce the burden SMV OR 101or count for the pair "c" and "o" (count - 23) ison the seP * ingohm by pWing only "podi prospects" to actually I=n than that for the pair of "ces" (count - 29) im-the matcher. Thig is eomplishm by aicuintim th teer plyin that. by this error metric, "c" and "o" are "closer"spewe dlctns btom the unkeowu sad each Igy ey, than the pair Of "e's" ais to each other. However, the errorand secng Soe Emy =NiMat wit the mAM disace pater for the Pair of "a's,'. which should be declared aathe bs pen Iect for a mach. li h atisike sj -, 1. the match. is composed of spenrLy distued pizels, while theamn buts emidss is conseed, and so fanth. The distmne am patt'I for the "o" ad "c" shows a dmns mode ofo"shc" sed to deterIne how "Clos" an uanknown is to a error Pixels corrsponding to the mbsing right segment of the
Vu 00111611 Is the "* block" distance defie by "a." One way to quantify the density of such a "node" asto form a summation in which the "local density" of every
Xblack pixel a merely the sum of all the pixels in its 3 X 3A(u.Ci 0(1re ,~rn) nelghborood if *9pixel is 1. and 0if the pixel isO. The
patars above labeled "weighated XOR error" JaIM beenwhMn 1c(1 is the 1h fetr of the =nMate, Ig(I) Is the calmiintsd In this saner. 16ot that by this eiterion theft I -- of do uk IN1- deinotes the absolut value, inmainfd amani inmate that the pair "c" and "o" aneml
mu(, C) is the mumm betswu the ushpow andgmist, aepanutmi (wat 13 1) then ma the Veit of "'e" (Count - 73).Mg Nis th -~ o lawmn.ft the taanplt matcher, the weighted wXCusmV OR swr
Is osapowd for ins ftestiee shift of a pair of patterana. TrsePAkN AMA Mltb seuig to bmizommul Mgd vertical. sluse pixel shift of
The aupis - seathartormsacorngerisma between the I~ ~s The inmusm ar a then compaed to aPwd~i of a dscied symbol and a libray psMgyp symbo. 111101611111 is 0911W t0 detrm whether Or not A match
Ciae I a di~msoal wm pata m F d by skoWi be declared. The value of the threshold is a non-
A(CR) ws C-1,2,'-.Ne Mg RUI2,'"-Njt. A Hese function of the symbol's block oomLand aobtinedaseI mples safhe alcularns the , - W be. bY se empbleay mindlook-up table.
towsf a 9 ol vector a -eI AM,.A) sad A(C,A) by L o y-oa" th9 1111er0 of Vplure elaMat (pi14e1) for wIs A ~ u seemeA(C, X) Mga I(C. A) ier. This UeXCWV OR efor is ds- A Aed else library is ased in the CSM system. The firsan hashed character and its faste vector occupy the first
X 0 XbwY dot The subsequent library slots a occusad bySo A! ,A 0BC X 2 these blooked duetsam for which so mnatch a found. bn
I* A(C,*)O5(CR)G(ow to is pi s o the thu o overnoftoig a sgoringspINI is ampesI to ack doe II of the librry
@ imarduos t n ameba onKV OIL us ONO .NI Whuend Iemy Is edo thes umsemd prowyp
3-3
- . -
PRATT or mL.: COMBDINED SYMBOL MATCHING FACSIMILE 769
I 11111 I I WILL1 I I
is 11111 I I 11112111ll ll l I tillI 221
MimH A OTnela An ~ tad MR1 term(b0Ie23 mti
ft. 4. Example of amuV3 as pettern for c md 0.
I III I~ 1 111111 1I1 333332 I
I u1 ILL I I Ilili II 1 1 II 3 I2
1 L 1121 I I I I 1 1 21111111 1 II 2 1
I in IL1U1U I luuuI 1 u 1111 2312223s II 11121322U I IU1.1.lU LI 1 II Is I
I 1 I .2IL I I I I I I I'
tI ll I12 I I I I amI
I 11uu1 11111 111111 1 1 1 11 231I lu I 111111 III 1 1I 232 1
ftfm A 1AIM! No I9tih4 3K tea
Ji.S. UZaple of naaM a& patt., for pa& Of a's.
ddddddddddddddddddddddddd
nnannnannnnnnnnnannnnngnnn
mmmminmmmimmmmmmmm
ppppppppppppppppppppppp'auuuuuuuuuuuuuuuuuuuuuuuGGGGTTTTvvvyvvYvrrrrrrrrrrrrrrrrrrrrrrrrrr
ftea. Peth Mbwaf plot of CCITT no. 4.
*...........................
*~ * to &
ft 1 Fardd 16my pies of CCITZ m. 7. (SyMbWolWddng pinfor-msn oeaWM 4WM a 161.)
to bumped an of the IIbumay ad z eI by he amw pinto- Pig. 6 and 7 ontalin Partia ibrazy plots of isolte symbolsqpe At dth e mw, dw -mgl do Ubm, mid the samei how two fmamimie docummna, one a French jownal arudslwaf Wonmm vm bWs to a qainbvo-- ' s wit (CCITT 44). and the othe a Japams WanWa docu~llmnt
*umm.m Wft& kwy otNd4mm, thesasm (CCITT On) The O b m am the bas bs the 5Mt solate po-Wum 0m OW 00W Pm~e a~ 4 Vmhe lot,e symbo, and Al bowing smbok ,ews.at mache
4 ~N am.~ a in*.to the posepe
8-4 * * t . . . . .....
I- 190PROCEEDIP4GS OF THE MEE,. VOL. 68. NO. 7. JULY 1960
h.~~~~S r. A njgg2 5 i
e . h - .1.- gnn
ata- nw . son 00Urwn & Is n 1 s
do 11rn P=.mama waf do Poet a4a-m.. an
in. spoo do i~ia am Nr. n am paM/' rjpf rono aS in for i
di a ft~ d g, n a . ro di mb. o inCO Id.d, 1W@Q do woit ma po %"inpin d.pd.11
OR) IGINAL
V d
(0) RESIDUE ,a
I1.L Nsmde sedes of CC1?IM 4.m sad i es.
A. frotoype Coding cdude eXCePtoily Warg Or excptionally small alphanumericAfter a symbol has been Mlocked, a decisn threshold is duchmcw, segments Of company logOs, and segments ofapplied to each plOtoty element of the libruy that ba handwnten scnwt In the CSM systemn, these patterns arepassed the Screening tat Vf a watch Is Indicaed, oy th rejected by the symnbol blocker, and then left behind as amatching library 10D ad horizontal locatLon with N P It o adue to be coded by two-dlmeamonml run4engh coding.dig PMeJOea symbol ase cods4. OthewjW the binery p-te- ftg 8 presets a blow up of a sectIon of a kodmle docu-
o0f the blocked symbol is usinlnted doesn Thm the sybol MUMt (ccnTR *4) lad its CON05!Wpoding residue.width, sybol bs* an hodsod location, in adasn ft ComeePtusfl, the CS yse could emnploy any type ofbt placd inth pRMwY mPe omdNMLd r14ogtcoing mtod freduscdig Tmdolr n
avfWeai, about 30 peau Of the prtotMp code bite On be Patbifty With hftWISim 3adards. Considering then factore,011mdisted by scnning the prototype pushk in a fodda Modified wemion of the CCrrr twO4Wdmem&Ina run4ength"bl"ke weave" Sequce and applyin g oe~hsi lb~au cdino algod"h has been selected for teresidue coder. ByHoffman coins of the n leagdu The dhedwua . of tW@ ihibns the symbol marching proes, the CSM coder wilWPproeg are addkknomi mplemuatdg omplexity and poe a001UtaORY b161"t to a pun residue coder, which can beOles lkw of 1 ite sre YlidVOINzS0210 whes a chanel wr JUG& 5Zxtl) cosr3etble with the CCrTF standard.ouu.The biawy coding appcoach bae been adopted for ahig~prfomeceversion of the Cat fanduile coder, and t A Tmuaium*on Code
fode r4ength coin 4mthod is used for a veq-hl01- The CSM facRASi coding SystM produces an asynchronousgarmanee veuoo. code that a dependent upon the contents of thle document toAeAWW CO&Wbe coded. Table I coINins a detaied spedcifatlon of the codeRi~in CEIIEU eleate and F*g.9 contain a state diapam defining thecoeIs3.MM dominenm, th s s Weak "Me fat dot do The code ots hma in tine apeeliakos have beew opti-not mat mni of rotompe ehesatus Ru N in- ined for a m rsmeono of a x 3 piablum
8-5
PRATT at . COMBINED SYMBOL MATCHING FACSIMILE F79zu1 Co
cm ~ ~ w cm WRDS
UNW Lw Sac 9 11DMc or IeMjMML Sm LOU.2s VO 3W K LO=
~W4 5W~. ?I L w a m seen. a m LOU
Pmm LM
UI LM flg. 7 N1mW am Or Lm =a
0. Upe STM10-r ew
Lwm a San 10 LX am 00m= "WNO
wim KO= COD %sn. am 9 orNA eu
UK= fin= 6.17 NORM~M 01 n azum WA=Lto MnW Kam ciseza amC -ODn
LWmm 01$ LOS
S4. lo~ am W&. i'o-onmmL u-ma. cm
W m =zur ~m mAlmome wfzO LOS uma
S"0 .ILISAT SOJYO
'C-LINE
C.U ~pL
5%IKSP3
.5no=
'SOL
NO MATC* S.C
KNIJ.A4malt
-C.'O
35g. 9. t es cede mtawswpin
I Extenadong f CSM Concept symbols. Isolated unknown symbols detected in the keyIs IL typica buggs letter scanned at 8 X 8 pixels/mm, pixel scanning pro.s that do not match a library entry can
* 4 about 40 percent Of the aompruNe code bits ean devoted to be placed in the Teddue for subsequent run-length coding.Mhe "12mNOW01 of Prototy sYmbolL IM Almtel of this The symbol matching process in the CSM system is notPOO=m Of the 4011mlll code on be 4111ete If the exct a match tolerance is parmiitted between symbols todomisem to be ,mble'me 1020we Ito a fiedast of accomodate perortabetom in symbol shape camse by thesymbols for 631811e, Cosder tYPewIe foat. kn th s e, eem2Wg ILosm Aa a consquence, in the basic CSM system,do the ke avid sher thd. an be p ipnosed with a v III, mix documt is not an exact ptxi-byplxe eIca
3792 PROCEEDINGS Of THE EEEE. VOL. 65. NO. 7. JULY 1980
* of the original document. Although symbol substitution (,
* errors are extremely rare, thern may be applications in which* exact coding is demanded. This mode of operation can be
accomodated in the CS14 system by a simple modification ofthe coder and decoder. At the coder, after a successful match,the EXCLUSIVE OR between the pair of matched symbols ii A
formed and placed in the residue for subsequent run4ength SIC.coding. At the decoder, the pixel arrays generated from re- Cc.~
* constructed symbols and reconstructed residue are combinedin an XCcLUSIVE OR fashion to correct for differences in thepair of matched symbols. In this manner, exact reproductionis achieved. However, the "overhead" associated with the Fag 10. LJne tracking.exact reproduction mode of operation can reduce the achiev-able compression ratio byua much as SO percent at 8X 8
Ypinel/mm resojution. where R represents the row index, C is the column index, Sdenotes the text line slope, and 0 is its offset. As characters
[HL SyNDOL RaCOGNiTioN MODE are encountered, they are assigned to the nearest straight lineThe CSM algorithm achieves facsimile data compression by representing a text line. The algorithm is as follows:
the matching of document symbols against a library of r-n 1) The coordinates (C, R) of the first encountered characterbole accumulated during the document scan. If a march oc- are used as a "seed" to start a cluster at S -0, 0- R.
curs th lirar inex s tansmtte rahertha th sybol 2) The (C, R) coordinates of the next character encounteredlibrary iate nde is anmitte athber etande the pembo are used to compute E -[R - S -C1 2 for the slope and offset
form symbol recognition by preloading the library with the thof h lser.o sls hnatrsodfragvncutrbinary psebn. Ptis b asirconceptican betextendedoto3) of eah lser.o sls hnatrsodfragvncutrbonary ph oe at penst syamymolre o lo h hrce sasge ota lse (next Line). If it is
font. Th codr cn thn opratein symolgrcoattio than the threshold for all clusters, the oldest clustermode in which only the ASCII cdsare trasmitted and all
* other document data such a a signature or logo are ignored. is dumped, and a new cluster is started.4) If the character was added to an existing cluster, the
* A. LUne Tracking values of slope and offset are updated by use of minimumi-
In the western world, printed matter is "read" from left to m sur io ehius
right and from top to bottom. Therefore, a symbol blocicing A. Handlng of Special Chars tenssystem that transmilts its output to a serial ASCII terminal A number of characters which consist of two "subcharacters"must do the sme. Howvever, the CSM algorithm extracts must be treated a special cases in the symbol-recognitioncharacters fromi the document being scanned in a totally dif- mode. This is because the blockerfmatcher would otherwiseferent fahion. As the line buffer scrofls through the pea fragment them into their constituent parts and give misleadingfrom top to bottom, the tallest of first encountered chawa- results. These characters are: (i), (j), (!), (?), (:), (;), (-), andten are removed from the document and processed through M.) After recognition of the two parts of the character, the
'.4the recognition algorithm. Thus characters emerge from the sytmwlchkifwoopabesmosaronoprCSM process in a sequence which would be totally incom- almost on top of each other. If so, the two symbols areprehensible if viewed in chronological sequence. In the con-ventional CSM facsimile transmission mode, this is of no con-wil merged into .Frampeto(so o fec tesequence, since characters are pleced in thmi appropriate wl emre noa()address locations regardless of thewr order of occurrence. Ina IV Comano RAXio Evju.nTmthe sesial symbol recognition mode, the tmansmitter wil W The CSU system has been extensively evaluated by com-algn each character as ASCII code, assemable the codes into puter simulation to optimize its performance and to determinelines, knserting bhmks, line feeds, catriage returns etc. and iscmrno ai ihrsett te oigmtostranmit the lines aerially to the receiver. For single spaced oritcopesnraowthepctoohecdngmhd.rotated documents, this "line tracking" Is more difficult tha A. Facsimile Mode Evaheationone wouradtemsine. iThL erlm inbina th atue of whpicg The CCITT document set of eight digitized documents ofthey chraceternitd lins Deein the suence in hich 200 X 200 line/in (8 X 8 pixels/mm) resolution, shown in
M ~theryol be tor mited t umn adrmea. A20steCa Fig. 11, has been used for evaluation of the CSM system in itstarsi may be sortd byCI tecmn istatesno A sinn facsmil mode of opeain. Tables 11 and ill conai listingsbenaer o tis a ed ASCe trm td ics th th Wifrm no of the compression ratios for each of the documents for thecharaeri lctatis neqed e rdter to 0 poety50Uc e high-performance and very-high-performance versions of thetqureve dsocumeth t. eurdi rert rpry~ CSII algorithm, respectively. These tables also contain the bit
The line-trackting algorithm Is based on a straight line fit of allocations for each of the code elements defined in Table I.doth key pixe coordinates of characters on a text line, as il- abl Vpesnsasumroomaio of the compresson
ha~~~~~~~nted ~~ ~ ~ ~ S Isyst1.Teemagtsieisdfnd aaetial with several other facsimile coding methods.ss The modified Huffman code is the CCITT adopted standard
R1SC-O (3) for onedlnmenalonal rue-length coding 121. The IBM code
B-7
,*-..-. ~.....*.**..*..,..*.*.*.*.%*.
614
PR.ATTo at.S: COMBINED SYMBOL MATCHING FACSIMILE 793
MEW
_X X
No. 1 No. 2 No. 3 No. I
-.- q
X..
No.1
famt I IuFORSIT
[7,READ code 181,adBOcd (91 are proposals for aThe compression factor obtained for this document for opera-CCITT standard employing two-dimensional rn-ength cod- tion of the CSM system in the symbol matching mode is aboutlag. These algorithms all provide for an end-of-ine code. All 257: 1 and for operation in the facsimile mode is atout 49: 1.of the algorithms in Table IV have been simulated and evali-
* ated on the same set of digttlzed documents scanned at the V. SYSTEM IMPLEMENTATION* Univesity of Hannover, Germany. The K factor indicates theO Although the CSM system is more complex to implement
number of lines in which the coder is operated in its two- thazi a conventional two-dimensional run-length coding sys-* dimensional mode before it revert to a one-dlmiensiona mode temn, with the advent of high-speed and relatively inexpensive
to limit the propagation of errors. -~ memory, discrete logic circuits, and microprocesors, imple-Comparison of the compremos performance of these aig- mentation complexity has ceased to be a deterrent to the
rithm Indicates that the CS(M mthds& outperform the run- development of high-perormance sysemis. A 100 X 100length coding technique. substantially for text-predominate lines/in (4 X 4 pixel/mm) facsimile coder using the CSMdocuments, and perform at about the Sam level as the best algorithm was introduced by Compression Labs, Inc. of'
%of the two-dimenajona run-length coding methods for graphics- Cupertino, CA, in Fall 1978. This unit utilime a inicroproces-predominate documents. sor to implement the algorithm for transmisaion at sub-
minute page rates. A discrete logic implementation of theA Symbol Rcogn iVAn Mode Zvlmo CSM algorithmn is being developed by Compression Labs for
The symbol recognition mode sper has beow tested with transmision rate of lessthan Sa for a200 X200 lnestinpage.son- of data, eaub Containing 1000 wimples of one of the
86 symbols of the Courier 10 font. In these tests, no mis- VI. SUMMARYmatches occurred, sad only very badly damaged characters A new highi-performance method of facsimile data compres-
wege 12ject ins ansmlsfabsnaslte a t e ion, called CSM, has been introduced. The coding systemFig 12conainan xamle f abusnes leteranditsre-involves segmentation of a document into symbols, that are
contuction with the symbol matching coding mode of opera- coded by template matching, and into a residue of the ro-tdon. It should be noted that the tratoucted letter has been m*aine of the document, that a coded by two-dimensionalprintedI wih a dillaut font than the orlgual, however, the nmength coding. Computer evaluation Indicates that the
a' fogimat and specisg of the two letters an in basic apesest. compen factor for tetpedominate documents is about
794 PROCEEDINGS OF THE IEEE. VOL. 6. NO. 7. JULY 1960
TABLE 11
HIG-Pvoamumcz Coou SummaIy
1 2 3 4 5 4 7 3II anl? 499 337 ZSS7 1613 1700 2299 6134 1792
LIWS?3 603 M0 603 403 603 403 603 403
symet 2126 2123 2126 2128 2123 2126 2126 2121
C-1IDE 2497 335 4169 8747 4895 1980 11174 1.584
NAxflG 981 37 1291 4015 1754 330 2522 156
LZ6Xo 5935 168 7574 26292 10773 1253 9370 434
moSPO "11 46 2164 7512 3076 358 2320 124
Max:2 1330 130 2090 2590 2150 Is10 1120 940
SLZMO 24491 3404 47036 55360 44789 30534 40647L. 2611
S.DBLCOI. 3227 233 9341 34177 14055 2431 25492 1024
3UCO 16594 11323 53430 20752 42014 93347 19345 161.3
?as ,**1072 "646 132233 163309 127939 137273 4990S3 218253
RM??O 54.5 37.2 27.7 22.4 28.7 26.3 7.4 16.A
TABLE IIIVvr Hizo-upomupmc Cooun SummaRy
5' uocm
1 2 3 4 5 6 7 3
J. SPZ? 433 264 1431 1162 1152 1540 3979 936
LIUTI m 403 403 603 603 $03 603 403 603
-YI 2123 2123 2123 21211 21261 IM12 2126 2121 --
-KA 2497 335 4169 3747 495 1ose 11.74 15"
VA M *as 3 1231 403 1754 330 2522 154 6
*. 5965 14 7974 24192 10773 1253 9470 434
am 1IM 43 214 7512 3073 353 2320 124- - --
="isl 1330 030 2090 2566 2150 1510 11120 940 ----- -
J M 1644 2053 30929 27393 23613 20937 322700 16239
tuc 3227 233 9341 34177 14055 2431 25492 1024
rimm 10594 M123 53130 20752 42014 93347 19345 131353
§91~6641 9*66 11 0 146417 11115 126637 41235S 207521
I - m
42.3 37.8 31.3 2S.3 33.1 29.0 3.6 17.7
4 .*
a.
PRATr or . CON DINED SYMBOL MATCHING FACSIMILE 795
TABLE IVCompuam L'Aiio Po Comwo oe CCLTT Documeir SgT wrnh vAsious CoDiNG Az~aomsr
R-3 9-4 9-3 X- 2-3 2.V r.117a 1.5.1 25.7 35.0 24.1 19.1 25.7 32.6 37.2 37.5
3 5.7 1.3.6 165.3 13.2 1.6 13.3 i5.@ 27.7 31.5
4 5.3 6.6 7.1 6.6 7.2 6.5 7.0 22.4 25.3
5 I.5 1.2.7 14.7 1.2.3 1.4.1 1.2.4 1.4.3 25.7 33.1
-6- 1.8.2 1.9.0 25.1 17.6 22.5 1.5.1 23.5 26.5 29.0
7 4.5 6.1 6.7 6.1 6.6 6.1 6.6 7.4 5.9
1 7.9 1.15.1 20.2 1.2.9 1.5.5 U..0 185.2 1.6.5 1.7.7
mAut Us Ws7 ag. 15. 1975
Seaetale a"auSewas aw . opagm Saeen~m.~ "egr
of N SM sem oonmhm soth.. aeompleb" for s ft- *I &Im salom cmpr~essionrtiose aceptable for On5 PAII
U.- taeGhAIN.1 ease aO.eOm. Us flig k ad the Come. amajila t as ea mr. 'Me nowe 44S 0 of thVX-e will be sho. w swmms e as a.6. mpied SC OS PAR34m will be able tosoe eta leasnt also sople @0 tole
~ IS as owcleb mUbfemia wil I a asssis page Prior to eelowing ImAt Will quareato a trammimies,umv s lows hm is. @os5 d u w Woo pswa" onoa~iftUe uaoflw ma n Sh.5 onsas fa t" M alo. Tbs. ttranislaesi
"m' tSo olmb o am6h hni35 86dt Roo Im abo gala h62e shawaA mug a 3400 booh digit" Nodes far
ooupgomsm lbUSI 51 wt so I 11 up isU heee ano e e ases of Lres Sal up toes 11 be
domldkm so- Pagm so sabionii. dewafto an ougasom Iam own# Iftee Go I.soasis. ooeagnev uag ae.sM80" iS e ONma On SMap". vases, simme segs am maval smmaS the pbages. mhemo mem lesi caies are
4SiNod U&S" mia S 96 UM1. PW 1406 mass1fdi 9110masS delhi afts mine she 96 line Poe lots bemm&"ieg lutiee
am@ . Noos CsOW 9. "AMmm:.
vue, "Osedeft icemesiom
mov 4) ORIGINAL Nes(b) REPRODIJCTIONIL. 12. M0 ofS doosawsSwi S mo u e m CM Sf
3-10
77
79* PROCEEDINGS OF THE IE. VOL. a.NO. 7. JULY 1980
*twice that obtid With two'dMfimnga i"a llgg Coding (41 W. IL Pratt. W. Chum. ad C. Roadle, UBloft chmow codagc"and about the Same for SPaphiwpndoMjnat docuumnm is AUG Sac h*teO.0~ LIIUomm"WoU Cffs&,Wmin VOL,T11 CS Ymcnb prtdi u mo c .iPP. 222-228. Amg. 1970.* T h CS SY te M GSA be ~ ii~ d 1 a m y~ bo r e o i- IS! W . C hMM , J. L- DO WO . aed R. D . W d uipes. -C o m hi d sy maboltlon mode in which a document is coded by recopiumo of itsi ~~hisS-A now appwosc to famhd dmt ourseulos,' Isa1Iphanumic symbols. CompruMo ratio WmaMe than 250: 1 AUG loG POpgPAW Iammm Iqie mI. 19",
call beS~le on busnes etm i ti mode of opmw 16 m -. A&Is
W-3~ . L .A m W IomL PrU W R H.- fma~kmAl 6III R. L. Arm -1m hewe. amagna171, J. Lug' 1Mkmum &Mmu, AG. lu& t= C -muuTiwo Cf.,o pp.md . ,
Tae~kW"% W. K. fn. Wd. Newv'= ior n Ia"md Promm, 1979. 1979. ~ x~QXlMCA P .. 197S121 #L G. bl and D. hems, -Cm who. of leddnamiu 18 ftop0mI for drat mmm"d" of two b mIcdnin~ Oedt for heak ftmmifto of dog"mnm, a= 27V tes"Re.CITSuy ru IceLbgo o 2Ceunms.. VOL CaM 25. . 11, PP. 142S-1433, Norv. I97. propmed byP CCTJSuyapan.V 60iueso 2(31 IL K. Amaw mad 0. pM. -A imm frc achomft ail -s. 19 Pomm for oaptmL aodm~a1cdn r..frp
.4 @et femaaim m IMMA.saud teat" Am 7%m caupur., hHOW189 apPm41M." Rep. CCITT StUd Group XIV. aSaUlbuoaVOL C-A3 PP. 1374-1179,1N4m. 1974. A-. 77. p'apod by bdftb Poo Offle Mw. 1979.
.9L
L%%
04,
4
* II
