5
Fast Retrieval of Old Dropcap Images Mathieu Delalandre, Jean-Marc Ogier and Josep Llados IDoc Meeting Valencia, Spain 22th February 2007

Fast Retrieval of Old Dropcap Images Mathieu Delalandre, Jean-Marc Ogier and Josep Llados IDoc Meeting Valencia, Spain 22th February 2007

Embed Size (px)

Citation preview

Page 1: Fast Retrieval of Old Dropcap Images Mathieu Delalandre, Jean-Marc Ogier and Josep Llados IDoc Meeting Valencia, Spain 22th February 2007

Fast Retrieval of Old Dropcap Images

Mathieu Delalandre, Jean-Marc Ogier and Josep Llados

IDoc Meeting

Valencia, Spain

22th February 2007

Page 2: Fast Retrieval of Old Dropcap Images Mathieu Delalandre, Jean-Marc Ogier and Josep Llados IDoc Meeting Valencia, Spain 22th February 2007

Fast Retrieval of Old Dropcap Images

Dropcaps in old books of XV° and XVI° Problematics

(1) Historian people are interested to retrieve similar printings

Wood plug(bottom

view)

Plug 1 Plug 2 Plug 3

Printing 1

Printing 2

(2) Several large databases are available on the web, a system must retrieve in real-time in order to allow cross queries

DB

DB

DB

query

query

r1 r2 r3

r1 r2 r3

Page 3: Fast Retrieval of Old Dropcap Images Mathieu Delalandre, Jean-Marc Ogier and Josep Llados IDoc Meeting Valencia, Spain 22th February 2007

Level 1 : image sizes Level 2 : image densityLevel 3 : RLE comparison

2121

2121

,max,max vvuu

vvuud

Our system (1/2)Our key ideas

(2) To use different level of operator (from more speed to more accurate)

query

1st Level

2sd Level

Speed

Depth

(1) To use a Run Length Encoding (RLE) of Image

Compression rate/Dropcap

0,7

0,8

0,9

1

Dropcap

Co

mp

res

sio

n r

ate

0.75

0.950.8

8

x2 x2 x2

x1x1 x1

x2

line (y) image

1

3

1

2 4

5 6

7line (y+dy)

image 2

while x1 x2 handle image 1

while x2 x1 handle image 2

Fast Retrieval of Old Dropcap Images

Page 4: Fast Retrieval of Old Dropcap Images Mathieu Delalandre, Jean-Marc Ogier and Josep Llados IDoc Meeting Valencia, Spain 22th February 2007

Our system (2/2)

2

clusterthreshold

Distance curve

00,10,20,30,40,50,60,70,8

1 167 333 499 665 831 997 1163 1329 1495 1661 1827 1993

Dropcap

Dis

tan

ce

1

2if 1 - 2 < 0

push x, cluster

while 1 - 2 < 0

next

Fast Retrieval of Old Dropcap Images

To switch between levels we use the ‘elbow criteria’

Results

32.89 s903.62 sMax

10.00 s337.06 sMean

05.35 s176.67 sMin

Our system

Raster comparisonQuery times

(laptop 1.8GHZ)

250 to 350 dpiResolutions

UncompressCompression

TiffFormat

grayModel

279.7 MoSize

2038Files

Test database

around 30 timesmore faster

0.1947 0.2517 0.3485 0.3616 0.3819 0.4064

Same plug

Next plug

Query

0.4109 0.4209

Example of query result

Page 5: Fast Retrieval of Old Dropcap Images Mathieu Delalandre, Jean-Marc Ogier and Josep Llados IDoc Meeting Valencia, Spain 22th February 2007

Fast Retrieval of Old Dropcap ImagesWorks in progress …

BaseOur

Retrieve engine

control

display

retrieve

Labels

driven labelling

Bench1 Bench2 Bench2To produce

To use our system like a groundtruthing one to evaluate retrieval results

0

5

10

15

20

25

30

35

40

0

5

10

15

20

25

30

35

40

To use a run based signature as additional level to speed up again the process

Application perspectives

To implement the engine on BVH website in next months

http://www.bvh.univ-tours.fr/

Contacts in progress with the EVODIA company to develop a commercial software

http://www.evodia.fr/