Dr. John R. Jensen Department of Geography University of South Carolina Columbia, SC 29208

Dr. John R. JensenDr. John R. JensenDepartment of GeographyDepartment of Geography

University of South CarolinaUniversity of South CarolinaColumbia, SC 29208Columbia, SC 29208

Digital Image Processing Hardware Digital Image Processing Hardware and System Considerationsand System Considerations

Jensen, 2004Jensen, 2004

Digital remote sensor data are analyzed using a Digital remote sensor data are analyzed using a digital digital image processing systemimage processing system that consists of computer that consists of computer hardware and special-purpose image processing software. hardware and special-purpose image processing software. This lecture describes: This lecture describes:

• fundamental digital image processing system fundamental digital image processing system hardwarehardware characteristics, characteristics, • digital image processing digital image processing softwaresoftware (computer program) (computer program) requirementsrequirements, and, and• public and commercialpublic and commercial sourcessources of digital image of digital image processing hardware and software. processing hardware and software.

Image Processing System ConsiderationsImage Processing System Considerations


A digital image processing system should:A digital image processing system should:

• have a reasonable have a reasonable learning curvelearning curve and be and be easy to useeasy to use, , • have a have a reputation for producing accurate resultsreputation for producing accurate results (ideally the (ideally the

company has ISO certification), company has ISO certification), • produce the desired results in an appropriate formatproduce the desired results in an appropriate format (e.g., (e.g.,

map products in a standard cartographic data structure map products in a standard cartographic data structure compatible with most GIS), and compatible with most GIS), and

• be be within the department’s budgetwithin the department’s budget. .



Computer Systems Computer Systems and Peripheral and Peripheral Devices in A Devices in A

Typical Digital Typical Digital Image Processing Image Processing

LaboratoryLaboratory


Image Processing System Image Processing System Hardware /Software ConsiderationsHardware /Software Considerations


• Number and speed of Number and speed of Central Processing Unit(s)Central Processing Unit(s) (CPU) (CPU)

• Operating systemOperating system (e.g., Microsoft Windows; UNIX, Linux, (e.g., Microsoft Windows; UNIX, Linux, Macintosh) Macintosh)

• Amount of Amount of random access memoryrandom access memory (RAM) (RAM)

• Number of image analysts that can use the system at one timeNumber of image analysts that can use the system at one time and and mode of operationmode of operation (e.g., interactive or batch) (e.g., interactive or batch)

• SerialSerial or or parallel image processingparallel image processing

• Arithmetic coprocessorArithmetic coprocessor or or array processorarray processor

• Software compiler(s)Software compiler(s) (e.g., C (e.g., C++++, Visual Basic, Java), Visual Basic, Java)





Image Processing System Image Processing System Hardware /Software ConsiderationsHardware /Software Considerations


• Type of mass storage (e.g., hard disk, CD-ROM, DVD) and amount (e.g., gigabytes)

• Monitor display spatial resolution (e.g., 1024 768 pixels)

• Monitor color resolution (e.g., 24-bits of image processing video memory yields 16.7 million displayable colors)

• Input devices (e.g., optical-mechanical drum or flatbed scanners, area array digitizers)

• Output devices (e.g., CD-ROM, CD-RW, DVD-RW, film-writers, line plotters, dye sublimation printers)

• Networks (e.g., local area, wide area, Internet)

Central Processing UnitCentral Processing Unit


The The central processing unitcentral processing unit (CPU) is the computing part of (CPU) is the computing part of the computer. It consists of a control unit and an arithmetic the computer. It consists of a control unit and an arithmetic logic unit. The CPU performs:logic unit. The CPU performs:

• numerical numerical integerinteger and/or and/or floating point calculationsfloating point calculations, and, and• directs input directs input and and outputoutput from and to mass storage devices, from and to mass storage devices, color monitors, digitizers, plotters, etc.color monitors, digitizers, plotters, etc.

Central Processing UnitCentral Processing Unit

A CPU’s efficiency is often measured in terms of how many millions-of-instructions-per-second (MIPS) it can process, e.g., 500 MIPS.

It is also customary to describe a CPU in terms of the number of cycles it can process in 1 second measured in megahertz, e.g., 1000 Mhz (1 GHz).

Manufacturers market computers with CPUs faster than 4 GHz, and this speed will increase. The system bus connects the CPU with the main memory, managing data transfer and instructions between the two. Therefore, another important consideration when purchasing a computer is bus speed.

Moore’s LawMoore’s Law

In 1985, Gordon Moore was preparing a speech and made an observation. He realized that each new computer CPU contained roughly twice as much capacity as its predecessor and each CPU was released within 18 to 24 months of the previous chip. If this trend continued, he reasoned, computing power would rise exponentially over relatively brief periods of time. Moore’s law described a trend that has continued and is still remarkably accurate. It is the basis for many planners’ performance forecasts. MIPS has also increased logarithmically.

History of Intel History of Intel MicroprocessorsMicroprocessors


4004 Microprocessor 4004 Microprocessor used in the used in the

Busicom CalculatorBusicom Calculator

Busicom CalculatorBusicom Calculator

Intel Intel Pentium 4Pentium 4



Type of Computer: Type of Computer:

* * Personal ComputersPersonal Computers (32 to 64-bit CPU) (32 to 64-bit CPU)

* * WorkstationWorkstation ( (>> 64-bit CPU) 64-bit CPU)

* * MainframeMainframe ( (>> 64-bit CPU) 64-bit CPU)

Type of ComputerType of Computer


Personal computersPersonal computers (16- to 64-bit CPUs) are the workhorses (16- to 64-bit CPUs) are the workhorses of digital image processing and GIS analysis. Personal of digital image processing and GIS analysis. Personal computers are based on microprocessor technology where computers are based on microprocessor technology where the the entire CPU is placed on a single chipentire CPU is placed on a single chip. These inexpensive . These inexpensive complex-instruction-set-computers (complex-instruction-set-computers (CISCCISC) generally have ) generally have CPUs with 32- to 64-bit registers (word size) that can CPUs with 32- to 64-bit registers (word size) that can compute integer arithmetic expressions at greater clock compute integer arithmetic expressions at greater clock speeds and process significantly more MIPS than their speeds and process significantly more MIPS than their 1980s – 1990s 8-bit predecessors. The 32-bit CPUs can 1980s – 1990s 8-bit predecessors. The 32-bit CPUs can process four 8-bit bytes at a time and 64-bit CPUs can process four 8-bit bytes at a time and 64-bit CPUs can process eight bytes at a time.process eight bytes at a time.






WorkstationsWorkstations usually consist of a usually consist of a >>64-bit reduced-64-bit reduced-instruction-set-computer (instruction-set-computer (RISCRISC) CPU that can address more ) CPU that can address more random access memory than personal computers. The RISC random access memory than personal computers. The RISC chip is typically faster than the traditional CISC. RISC chip is typically faster than the traditional CISC. RISC workstations application software and hardware workstations application software and hardware maintenance costs are usually higher than personal maintenance costs are usually higher than personal computer-based image processing systems. The most computer-based image processing systems. The most common workstation operating systems are UNIX and common workstation operating systems are UNIX and various Microsoft Windows products.various Microsoft Windows products.



Mainframe computersMainframe computers ( (>>64-bit CPU) perform calculations 64-bit CPU) perform calculations more rapidly than PCs or workstations and able to support more rapidly than PCs or workstations and able to support hundreds of users simultaneously, especially parallel hundreds of users simultaneously, especially parallel mainframe computers such as a CRAY. This makes mainframe computers such as a CRAY. This makes mainframes ideal for intensive, CPU-dependent tasks (e.g., mainframes ideal for intensive, CPU-dependent tasks (e.g., image rectification, mosaicking, filtering, classification, image rectification, mosaicking, filtering, classification, hyperspectral image analysis, and GIS modeling). If desired, hyperspectral image analysis, and GIS modeling). If desired, the output from intensive mainframe processing can be the output from intensive mainframe processing can be passed to a workstation or personal computer for subsequent passed to a workstation or personal computer for subsequent less intensive, inexpensive processing. Mainframe computer less intensive, inexpensive processing. Mainframe computer systems are expensive to purchase and maintain. Mainframe systems are expensive to purchase and maintain. Mainframe applications software is more expensive. applications software is more expensive.



The The operating systemoperating system is the first program loaded into is the first program loaded into random access memory (RAM) when the computer is turned random access memory (RAM) when the computer is turned on. It controls the computer’s higher-order functions. The on. It controls the computer’s higher-order functions. The operating system kernel resides in memory at all times. The operating system kernel resides in memory at all times. The operating system provides the user interface and controls operating system provides the user interface and controls multitasking. It handles the input and output to the hard disk multitasking. It handles the input and output to the hard disk and all peripheral devices such as compact disks, scanners, and all peripheral devices such as compact disks, scanners, printers, plotters, and color displays. All digital image printers, plotters, and color displays. All digital image processing application programs must communicate with the processing application programs must communicate with the operating system. The operating system sets the protocols operating system. The operating system sets the protocols for the application programs that are executed by it. for the application programs that are executed by it.

Operating SystemOperating System


The difference between a single-user operating system and a The difference between a single-user operating system and a network operating system is the latter’s network operating system is the latter’s multi-user multi-user capabilitycapability..

• Microsoft Windows XPMicrosoft Windows XP (home edition) and the (home edition) and the Macintosh Macintosh OSOS are single-user operating systems designed for one are single-user operating systems designed for one person at a desktop computer working independently.person at a desktop computer working independently.

• Various Various Microsoft WindowsMicrosoft Windows, , UNIXUNIX, and , and LinuxLinux network network operating systems are designed to manage multiple user operating systems are designed to manage multiple user requests at the same time and complex network security.requests at the same time and complex network security.

Operating SystemsOperating Systems

Read Only Memory and Random Access MemoryRead Only Memory and Random Access Memory


Read-only memoryRead-only memory (ROM) retains information even after the (ROM) retains information even after the computer is shut down because power is supplied from a computer is shut down because power is supplied from a battery that must be replaced occasionally. Most computers battery that must be replaced occasionally. Most computers have sufficient ROM for digital image processing have sufficient ROM for digital image processing applications; therefore, it is not a serious consideration. applications; therefore, it is not a serious consideration.

Random access memoryRandom access memory (RAM) is the computer’s primary (RAM) is the computer’s primary temporarytemporary workspace. It requires power to maintain its workspace. It requires power to maintain its content. Therefore, all of the information that is temporarily content. Therefore, all of the information that is temporarily placed in RAM while the CPU is performing digital image placed in RAM while the CPU is performing digital image processing must be saved to a hard disk (or other media such processing must be saved to a hard disk (or other media such as a CD) before turning the computer off.as a CD) before turning the computer off.

Read Only Memory and Random Access MemoryRead Only Memory and Random Access Memory


Computers should have sufficient RAM for the operating Computers should have sufficient RAM for the operating system, image processing applications software, and any system, image processing applications software, and any remote sensor data that must be held in temporary memory remote sensor data that must be held in temporary memory while calculations are performed. Computers with 64-bit while calculations are performed. Computers with 64-bit CPUs can address more RAM than 32-bit machines CPUs can address more RAM than 32-bit machines

• It seems that one can never have too much RAM for image It seems that one can never have too much RAM for image processing applications. RAM prices continue to decline processing applications. RAM prices continue to decline while RAM speed continues to increase.while RAM speed continues to increase.


One of the best One of the best scientific visualization environmentsscientific visualization environments for the for the analysis of remote sensor data takes place when the analyst analysis of remote sensor data takes place when the analyst communicates with the digital image processing system communicates with the digital image processing system interactivelyinteractively using a point-and-clickusing a point-and-click graphical user graphical user interface interface (GUI)(GUI). Most sophisticated image processing . Most sophisticated image processing systems are now configured with a friendly GUI that allows systems are now configured with a friendly GUI that allows rapid display of images and the selection of important image rapid display of images and the selection of important image processing functions. processing functions.

Interactive Graphical User Interface (GUI)Interactive Graphical User Interface (GUI)


Several effective digital image processing graphical user interfaces include:

• ERDAS Imagine’s intuitive point-and-click icons, • Research System’s Environment for Visualizing Images (ENVI) hyperspectral data analysis interface,

• ER Mapper, • IDRISI, • ESRI ArcGIS Image Analyst, and • Adobe Photoshop.

Graphical User InterfaceGraphical User Interface

ENVI Interface ENVI Interface


ENVI Interface ENVI Interface


ERDAS Interface ERDAS Interface



Non-interactive, Non-interactive, batch processingbatch processing is of value for time- is of value for time-consuming processes such as image rectification, consuming processes such as image rectification, mosaicking, orthophoto creation, filtering, etc. mosaicking, orthophoto creation, filtering, etc.

• Batch processing frees up lab PCs or workstations during Batch processing frees up lab PCs or workstations during peak demand because the jobs can be stored and executed peak demand because the jobs can be stored and executed when the computer is idle (e.g., early morning hours). when the computer is idle (e.g., early morning hours).

• Batch processing can also be useful during peak hours Batch processing can also be useful during peak hours because it allows the analyst to set up a series of because it allows the analyst to set up a series of operations that can be executed in sequence without operations that can be executed in sequence without operator intervention. operator intervention.

• Digital image processing also can now be performed Digital image processing also can now be performed interactively over the Internet at selected sites.interactively over the Internet at selected sites.

Interactive versus Batch ProcessingInteractive versus Batch Processing


Digital image processing of remote sensing and related GIS Digital image processing of remote sensing and related GIS data requires substantial mass storage resources. Mass data requires substantial mass storage resources. Mass storage media should have:storage media should have:

• rapid access timerapid access time, , • longevitylongevity (i.e., last for a long time), and be (i.e., last for a long time), and be • inexpensiveinexpensive..

Storage and Archiving ConsiderationsStorage and Archiving Considerations

Serial and Parallel Image ProcessingSerial and Parallel Image Processing


It is possible to obtain PCs, workstations, and mainframe It is possible to obtain PCs, workstations, and mainframe computers that have multiple CPUs that operate computers that have multiple CPUs that operate concurrently. Specially written concurrently. Specially written parallel processing softwareparallel processing software can can parseparse (distribute) the remote sensor data to specific (distribute) the remote sensor data to specific CPUs to perform digital image processing. This can be much CPUs to perform digital image processing. This can be much more efficient than processing the data serially. more efficient than processing the data serially.

Serial Versus Parallel Processing

Jensen, 2004

• Requires more than one CPU

• Requires software that can parse (distribute) the digital image processing to the various CPUs by

- task, and/or

- line, and/or

- column.



Consider performing a per-pixel Consider performing a per-pixel classification on a 1024 row by 1024 classification on a 1024 row by 1024 column remote sensing dataset. In the column remote sensing dataset. In the first example, each pixel is classified first example, each pixel is classified by passing the spectral data to the by passing the spectral data to the CPU and then progressing to the next CPU and then progressing to the next pixel. This is pixel. This is serialserial processing. processing. Conversely, suppose that instead Conversely, suppose that instead of just one CPU we had 1024 CPUs. of just one CPU we had 1024 CPUs. In this case the class of each of the In this case the class of each of the 1024 pixels in the row could be 1024 pixels in the row could be determined using 1024 separate determined using 1024 separate CPUs. The CPUs. The parallelparallel image processing image processing would classify the line of data about would classify the line of data about 1024 times faster than would 1024 times faster than would processing it serially. processing it serially.



Each of the 1024 CPUs could also be allocated an entire row Each of the 1024 CPUs could also be allocated an entire row of the dataset. Finally, each of the CPUs could be allocated a of the dataset. Finally, each of the CPUs could be allocated a separate band if desired. For example, if 224 bands of separate band if desired. For example, if 224 bands of AVIRIS hyperspectral data were available, 224 of the 1024 AVIRIS hyperspectral data were available, 224 of the 1024 processors could be allocated to evaluate the 224 brightness processors could be allocated to evaluate the 224 brightness values associated with each individual pixel with 800 values associated with each individual pixel with 800 additional CPUs available for other tasks.additional CPUs available for other tasks.


Serial versus Parallel Digital Image Processing to Perform Per-pixel Classification


A computer software compiler translates instructions programmed in a high-level language such as C++ or Visual Basic into machine language that the CPU can understand. A compiler usually generates assembly language first and then translates the assembly language into machine language. The compilers most often used in the development of digital image processing software are C++, Assembler, and Visual Basic. Many digital image processing systems provide a toolkit that programmers can use to compile their own digital image processing algorithms (e.g., ERDAS, ER Mapper, ENVI). The toolkit consists of subroutines that perform very specific tasks such as reading a line of image data into RAM or modifying a color look-up table to change the color of a pixel (RGB) on the screen.

CompilerCompiler


It is often useful for remote sensing analysts to program in It is often useful for remote sensing analysts to program in one of the high-level languages just listed. Very seldom will one of the high-level languages just listed. Very seldom will a single digital image processing system perform all of the a single digital image processing system perform all of the functions needed for a given project. Therefore, the ability functions needed for a given project. Therefore, the ability to modify existing software or integrate newly developed to modify existing software or integrate newly developed algorithms with the existing software is important. algorithms with the existing software is important.

CompilerCompiler

Digital remote sensor data (and ancillary raster GIS data) are often stored in a matrix band sequential (BSQ) format in which each spectral band of imagery (or GIS data) is stored as an individual file. Each picture element of each band is typically represented in the computer by a single 8-bit byte with values from 0 to 255.

The best way to make brightness values rapidly available to the computer is to place the data on a hard disk, CD-ROM, DVD, or DVD-RAM where each pixel of the data matrix may be accessed at random (not serially) and at great speed (e.g., within microseconds). The cost of hard disk, CD-ROM, or DVD storage per gigabyte continues to decline.

Rapid Access Mass StorageRapid Access Mass Storage



It is common for digital image processing laboratories to It is common for digital image processing laboratories to have gigabytes of hard-disk mass storage associated with have gigabytes of hard-disk mass storage associated with each workstation. Many image processing labs now use each workstation. Many image processing labs now use RAID (redundant arrays of inexpensive hard disks) RAID (redundant arrays of inexpensive hard disks) technology in which two or more drives working together technology in which two or more drives working together provide increased performance and various levels of error provide increased performance and various levels of error recovery and fault tolerance. Other storage media, such as recovery and fault tolerance. Other storage media, such as magnetic tapes, are usually too slow for real-time image magnetic tapes, are usually too slow for real-time image retrieval, manipulation, and storage because they do not retrieval, manipulation, and storage because they do not allow random access of data. However, given their large allow random access of data. However, given their large storage capacity, they remain a cost-effective way to archive storage capacity, they remain a cost-effective way to archive digital remote sensor data.digital remote sensor data.



Companies are now developing new mass storage Companies are now developing new mass storage technologies based on technologies based on atomic resolution storageatomic resolution storage (ARS), (ARS), which holds the promise of storage densities of close to 1 which holds the promise of storage densities of close to 1 terabit per square inch—the equivalent of nearly 50 DVDs terabit per square inch—the equivalent of nearly 50 DVDs on something the size of a credit card. The technology uses on something the size of a credit card. The technology uses microscopic probes less than one-thousandth the width of a microscopic probes less than one-thousandth the width of a human hair. When the probes are brought near a conducting human hair. When the probes are brought near a conducting material, electrons write data on the surface. The same material, electrons write data on the surface. The same probes can detect and retrieve data and can be used to write probes can detect and retrieve data and can be used to write over old data.over old data.



Storing remote sensor data is no trivial matter.Storing remote sensor data is no trivial matter. Significant Significant sums of money are spent purchasing remote sensor data by sums of money are spent purchasing remote sensor data by commercial companies, natural resource agencies, and commercial companies, natural resource agencies, and universities. Unfortunately, most of the time not enough universities. Unfortunately, most of the time not enough attention is given to how the expensive data are stored or attention is given to how the expensive data are stored or archived to protect the long-term investment. The diagram archived to protect the long-term investment. The diagram depicts several types of analog and digital remote sensor depicts several types of analog and digital remote sensor data mass storage devices and the average time to physical data mass storage devices and the average time to physical obsolescence, that is, when the media begin to deteriorate obsolescence, that is, when the media begin to deteriorate and information is lost. and information is lost.

Archiving Considerations and LongevityArchiving Considerations and Longevity

Potential Potential Longevity of Longevity of

Remote Remote Sensor Data Sensor Data

Storage Storage MediaMedia



• Properly exposed, washed, and fixed Properly exposed, washed, and fixed analog black & white analog black & white aerial photography negativesaerial photography negatives have considerable longevity, have considerable longevity, often more than 100 years. often more than 100 years.

• Color negativesColor negatives with their respective dye layers have with their respective dye layers have longevity, but not as much as the black-and-white longevity, but not as much as the black-and-white negatives.negatives.

• Black & white paper printsBlack & white paper prints have greater longevity than have greater longevity than color prints (Kodak, 1995). color prints (Kodak, 1995).

• Hard and floppy magnetic disksHard and floppy magnetic disks have relatively short have relatively short longevity, often less than 20 years. longevity, often less than 20 years.

• Magnetic tape mediaMagnetic tape media (e.g., 3/4-in. tape, 8-mm tape) can (e.g., 3/4-in. tape, 8-mm tape) can become unreadable within 10 to 15 years if not rewound become unreadable within 10 to 15 years if not rewound and properly stored in a cool, dry environment.and properly stored in a cool, dry environment.


Potential Potential Longevity of Longevity of

Remote Remote Sensor Data Sensor Data

Storage Storage MediaMedia



Optical disksOptical disks can now be written to, read, and can now be written to, read, and written over again at relatively high speeds and can store written over again at relatively high speeds and can store much more data than other portable media such as floppy much more data than other portable media such as floppy disks. The technology used in rewriteable optical systems is disks. The technology used in rewriteable optical systems is magneto-opticsmagneto-optics, where data is recorded magnetically like , where data is recorded magnetically like disks and tapes, but the bits are much smaller because a disks and tapes, but the bits are much smaller because a laser is used to etch the bit. The laser heats the bit to 150 °C, laser is used to etch the bit. The laser heats the bit to 150 °C, at which temperature the bit is realigned when subjected to a at which temperature the bit is realigned when subjected to a magnetic field. To record new data, existing bits must first magnetic field. To record new data, existing bits must first be set to zero. be set to zero.



Only the Only the optical diskoptical disk provides relatively long-term storage provides relatively long-term storage potential (>100 years). In addition, optical disks store large potential (>100 years). In addition, optical disks store large volumes of data on relatively small media. Advances in volumes of data on relatively small media. Advances in optical compact disc (CD) technology promise to increase optical compact disc (CD) technology promise to increase the storage capacity to > 17 Gb using new rewriteable the storage capacity to > 17 Gb using new rewriteable digital video disc (DVD) technology. In most remote digital video disc (DVD) technology. In most remote sensing laboratories, rewritable CD-RWs or DVD-RWs sensing laboratories, rewritable CD-RWs or DVD-RWs have supplanted tapes as the backup system of choice. DVD have supplanted tapes as the backup system of choice. DVD drives are backwards compatible and can read data from drives are backwards compatible and can read data from CDs.CDs.



It is important to remember when archiving remote sensor It is important to remember when archiving remote sensor data that sometimes it is the loss of the data that sometimes it is the loss of the

• read-write softwareread-write software and/or and/or • read-write hardwareread-write hardware (the drive mechanism and heads) (the drive mechanism and heads)

that is the problem and not the digital media itself. that is the problem and not the digital media itself. Therefore, as new computers are purchased it is a good idea Therefore, as new computers are purchased it is a good idea to set aside a single computer system that is representative to set aside a single computer system that is representative of a certain computer era so that one can always read any of a certain computer era so that one can always read any data stored on data stored on archivedarchived mass storage media. mass storage media.


Computer Display Spatial and Color Resolution


The display of remote sensor data on a computer screen is The display of remote sensor data on a computer screen is one of the most fundamental elements of digital image one of the most fundamental elements of digital image analysis. Careful selection of the computer display analysis. Careful selection of the computer display characteristics will provide the optimum visual image characteristics will provide the optimum visual image analysis environment for the human interpreter. The two analysis environment for the human interpreter. The two most important characteristics are computer :most important characteristics are computer :

• display spatial resolutiondisplay spatial resolution, and , and • color resolutioncolor resolution..

Computer Screen Display Resolution


The image processing system should be able to display at least 1024 rows by 1024 columns on the computer screen at one time. This allows larger geographic areas to be examined and places the terrain of interest in its regional context. Most Earth scientists prefer this regional perspective when performing terrain analysis using remote sensor data. Furthermore, it is disconcerting to have to analyze four 512 512 images when a single 1024 1024 display provides the information at a glance. An ideal screen display resolution is 1600 1200 pixels.





Computer Screen Color Resolution


The computer screen color resolution is the number of gray-scale tones or colors (e.g., 256) that can be displayed on a CRT monitor at one time out of a palette of available colors (e.g., 16.7 million). For many applications, such as high-contrast black-and-white linework cartography, only 1 bit of color is required [i.e., either the line is black or white (0 or l)]. For more sophisticated computer graphics for which many shades of gray or color combinations are required, up to 8 bits (or 256 colors) may be required. Most thematic mapping and GIS applications may be performed quite well by systems that display just 64 user-selectable colors out of a palette of 256 colors.



The analysis and display of remote sensor image data generally requires much higher CRT screen color resolution than cartographic and GIS applications. For example, most relatively sophisticated digital image processing systems can display a tremendous number of unique colors (e.g., 16.7 million) from a large color palette (e.g., 16.7 million). The primary reason for these color requirements is that image analysts must often display a composite of several images at one time on a CRT. This process is called color compositing.







To display a To display a color-infrared imagecolor-infrared image of Landsat Thematic of Landsat Thematic Mapper data it is necessary to composite three separate 8-bit Mapper data it is necessary to composite three separate 8-bit images [e.g., green band (TM 2 = 0.52 to 0.60 µm), red band images [e.g., green band (TM 2 = 0.52 to 0.60 µm), red band (TM 3 = 0.63 to 0.69 µm), reflective infrared band (TM 4 = (TM 3 = 0.63 to 0.69 µm), reflective infrared band (TM 4 = 0.76 to 0.90 µm)]. To obtain a 0.76 to 0.90 µm)]. To obtain a true-colortrue-color composite image composite image that provides every possible color combination for the three that provides every possible color combination for the three 8-bit images requires that 28-bit images requires that 22424 colors (16,777,216) be colors (16,777,216) be available in the palette. Such true-color systems are available in the palette. Such true-color systems are relatively expensive because every pixel location must be relatively expensive because every pixel location must be bitmappedbitmapped. . This means that there must be a specific location This means that there must be a specific location in memory that keeps track of the exact blue, green, and red in memory that keeps track of the exact blue, green, and red color value for every pixel. This requires substantial color value for every pixel. This requires substantial computer memory which are usually collected in what is computer memory which are usually collected in what is called an called an image processorimage processor.. Given the availability of image Given the availability of image processor memory, the question is: what is adequate color processor memory, the question is: what is adequate color resolution?resolution?



Generally, 4096 carefully selected colors out of a very large Generally, 4096 carefully selected colors out of a very large palette (e.g., 16.7 million) appears to be the minimum palette (e.g., 16.7 million) appears to be the minimum acceptable for the creation of remote sensing color acceptable for the creation of remote sensing color composites. This provides 12 bits of color, with 4 bits composites. This provides 12 bits of color, with 4 bits available for each of the blue, green, and red image planes. available for each of the blue, green, and red image planes. For image processing applications other than compositing For image processing applications other than compositing (e.g., black & white image display, color density slicing, (e.g., black & white image display, color density slicing, pattern recognition classification), the 4096 available colors pattern recognition classification), the 4096 available colors and large color palette are more than adequate. However, the and large color palette are more than adequate. However, the larger the palette and the greater the number of displayable larger the palette and the greater the number of displayable colors at one time, the better the representation of the remote colors at one time, the better the representation of the remote sensor data on the CRT screen for visual analysis. sensor data on the CRT screen for visual analysis.

Important Image Processing Functions


Many of the most important functions performed using digital image processing systems are summarized in Table 3-4. Personal computers now have the computing power to perform each of these functions.

Image Processing System FunctionsImage Processing System Functions

Jensen, 2004

* Preprocessing (Radiometric and Geometric)* Preprocessing (Radiometric and Geometric)

* Display and Enhancement* Display and Enhancement

* Information Extraction* Information Extraction

* Photogrammetric Information Extraction* Photogrammetric Information Extraction

* Metadata and Image/Map Lineage Documentation * Metadata and Image/Map Lineage Documentation

* Image and Map Cartographic Composition* Image and Map Cartographic Composition

* Geographic Information Systems (GIS)* Geographic Information Systems (GIS)

* Integrated Image Processing and GIS* Integrated Image Processing and GIS

* Utilities* Utilities



It is not good for remotely sensed data to be analyzed in a It is not good for remotely sensed data to be analyzed in a vacuum. vacuum. Remote sensing information fulfills its promise best Remote sensing information fulfills its promise best when used in conjunction with ancillary data (e.g., soils, when used in conjunction with ancillary data (e.g., soils, elevation, and slope) stored in a geographic information elevation, and slope) stored in a geographic information system (GIS).system (GIS). The ideal system should be able to process the The ideal system should be able to process the digital remote sensor data and perform any necessary GIS digital remote sensor data and perform any necessary GIS processing. It is not efficient to exit the digital image processing. It is not efficient to exit the digital image processing system, log into a GIS system, perform a required processing system, log into a GIS system, perform a required GIS function, and then take the output of the procedure back GIS function, and then take the output of the procedure back into the digital image processing system for further analysis. into the digital image processing system for further analysis. Integrated systems perform both digital image processing Integrated systems perform both digital image processing and GIS functions and consider map data as image data (or and GIS functions and consider map data as image data (or vice versa) and operate on them accordingly. vice versa) and operate on them accordingly.



Most digital image processing systems have limitations.Most digital image processing systems have limitations. For For example:example:• Most systems can perform multispectral classification on a Most systems can perform multispectral classification on a few bands of imagery, but only a few systems can perform few bands of imagery, but only a few systems can perform hyperspectral analysis on hundreds of bands of imagery.hyperspectral analysis on hundreds of bands of imagery.• Only a few systems can perform soft-copy Only a few systems can perform soft-copy photogrammetric operations on stereoscopic imagery photogrammetric operations on stereoscopic imagery displayed on the CRT screen to generate digital orthophotos displayed on the CRT screen to generate digital orthophotos and digital elevation models. and digital elevation models. • Only a few systems incorporate expert systems, neural Only a few systems incorporate expert systems, neural networks or fuzzy logic (e.g., ERDAS Imagine, ER Mapper). networks or fuzzy logic (e.g., ERDAS Imagine, ER Mapper). • Few provide image lineage (genealogy) data about the Few provide image lineage (genealogy) data about the processing applied to each image. processing applied to each image.

Commercial and Public Digital Image Processing Systems


Commercial firms actively market digital image processing systems. Some companies provide only the software, others provide both proprietary hardware and software.

Public agencies (e.g., NASA, NOAA, and the Bureau of Land Management) and universities (e.g., Purdue University) have developed digital image processing software. Some of these public systems are available at minimal cost.


Jensen, 2004Jensen, 2004Selected Commercial and Public Digital Image Processing Systems


Selected Commercial and Public Digital Image Processing Systems

Jensen, 2004Jensen, 2004Selected Commercial and Public Digital Image Processing Systems

Major Commercial Major Commercial Digital Image Processing SystemsDigital Image Processing Systems


- ERDASERDAS

- Leica Photogrammetry SuiteLeica Photogrammetry Suite

- ENVI- ENVI

- IDRISI- IDRISI

- ER Mapper- ER Mapper

- PCI GeomaticaPCI Geomatica

- eeCognitionCognition


- GRASS- GRASS

- MultiSpec (LARS Purdue University)- MultiSpec (LARS Purdue University)

- C-CoastC-Coast

- Adobe PhotoshopAdobe Photoshop

Major Public Major Public Digital Image Processing SystemsDigital Image Processing Systems

Sources of Digital Image Processing Systems


ACORN, Atmospheric CORrection Now, www.aigllc.com/acorn/ intro.aspAGIS Software, www.agismap.comApplied Analysis Inc., Subpixel Processing, www.discover-aai.comArcGIS Feature Analyst; www.featureanalyst.comATCOR2, www.geosystems.de/atcor/atcor2.htmlAUTOCAD, Autodesk, Inc., usa.autodesk.comBAE Systems SOCET Set, www.socetset.comBlue Marble Geographics, www.bluemarblegeo.com. C-Coast, http://coastwatch.noaa.gov/cw_ccoast.htmlCosmic, www.openchannelfoundation.org/cosmicDIMPLE, www.process.com.au/AboutDIMPLE.shtmlDragon, Goldin-Rudahl Systems, www.goldin-rudahl.comEarthView, Atlantis Scientific Systems, www.pcigeomatics.comEIDETIC Earthscope, www.eidetic.bc.ca/~eidetic/es1.htmENVI, Research Systems, Inc., www.rsinc.comELAS (DIPIX, Datastar), http://technology.ssc.nasa.gov/PDFs/ SSC-00001_SS_NTTS.pdfERDAS Imagine, www.erdas.comER Mapper, www.ermapper.comFullPixelSearch, www.themesh.com/elink13.htmlGlobal Lab, Data Translation, 100 Locke Dr., Marlboro, MA 01752-1192GRASS, http://grass.itc.itIDRISI, Clarke University, www.clarklabs.orgImagePro, www.i-cubeinc.com/software.htmIntelligent Library System, Lockheed Martin, www.lmils.comIntergraph, www.intergraph.com

Sources of Digital Image Processing Systems


MapInfo, www.mapinfo.comMacSadie, www.ece.arizona.edu/~dial/base_files/software/ MacSadie1.2.htmlMrSID, LizardTech, www.lizardtech.comMultiSpec, www.ece.purdue.edu/~biehl/MultiSpec/.NIH-Image, http://rsb.info.nih.gov/nih-imageNOeSYS, www.rsinc.com/NOESYS/index.cfmPCI, www.pcigeomatics.comPHOTOSHOP, www.adobe.comRemoteView, www.sensor.com/remoteview.htmlR-WEL Inc., www.rwel.comTNTmips, MicroImages, www.microimages.comVISILOG, www.norpix.com/visilog.htmXV image viewer, www.trilon.com/xv

Digital Image Processing and the Digital Image Processing and the National Spatial Data InfrastructureNational Spatial Data Infrastructure


Laypersons and scientists who use remote sensing data or Laypersons and scientists who use remote sensing data or share products derived from remotely sensed data should be share products derived from remotely sensed data should be aware of aware of spatial data standardsspatial data standards developed by the Federal developed by the Federal Geographic Data CommitteeGeographic Data Committee (FGDC). The FGDC is an (FGDC). The FGDC is an interagency committee of representatives from the Executive interagency committee of representatives from the Executive Office of the President, the Cabinet, and independent Office of the President, the Cabinet, and independent agencies. The FGDC is developing the agencies. The FGDC is developing the National Spatial Data National Spatial Data InfrastructureInfrastructure (NSDI) in cooperation with organizations from (NSDI) in cooperation with organizations from state, local and tribal governments, academics, and the state, local and tribal governments, academics, and the private sector. The NSDI encompasses policies, standards, private sector. The NSDI encompasses policies, standards, and procedures for organizations to cooperatively produce and procedures for organizations to cooperatively produce and share geographic data.and share geographic data.

Digital Image Processing and the Digital Image Processing and the National Spatial Data InfrastructureNational Spatial Data Infrastructure


The American National Standards Institute’s (ANSI) The American National Standards Institute’s (ANSI) Spatial Spatial Data Transfer StandardData Transfer Standard (SDTS) is a mechanism for (SDTS) is a mechanism for archiving and transferring spatial data (including metadata) archiving and transferring spatial data (including metadata) between dissimilar computer systems. The SDTS specifies between dissimilar computer systems. The SDTS specifies exchange constructs, such as format, structure, and content, exchange constructs, such as format, structure, and content, for spatially referenced vector and raster (including gridded) for spatially referenced vector and raster (including gridded) data. Actual use of SDTS to transfer spatial data is carried data. Actual use of SDTS to transfer spatial data is carried out through its profiles. The out through its profiles. The FGDC FGDC Raster ProfileRaster Profile standard is standard is of particular interest because it provides specifications for of particular interest because it provides specifications for transferring spatial datasets in which features or images are transferring spatial datasets in which features or images are represented in raster or gridded form, such as digital represented in raster or gridded form, such as digital elevation models, digital orthophoto quarter quads (DOQQ), elevation models, digital orthophoto quarter quads (DOQQ), and digital satellite imagery.and digital satellite imagery.

Remote Sensing Data FormatsRemote Sensing Data Formats


* Band interleaved by line (BIL)* Band interleaved by line (BIL)

* Band interleaved by pixel (PIP)* Band interleaved by pixel (PIP)

* Band sequential (BSQ)* Band sequential (BSQ)

* Run-length encoding* Run-length encoding

Documents

Dr. John R. Jensen Department of Geography University of South Carolina Columbia, SC 29208