Computer Hardware for Fast Image Processing

The tremendous progress of computer technology in the past 20 years has brought digital image processing to the desk of every scientist and engineer. For a general-purpose computer to be useful for image process- ing, four key demands must be met: high-resolution image display, suf- fi cient memory transfer bandwidth, suffi cient storage space, and suffi - cient computing power. In all four areas, a critical level of performance has been reached that makes it possible to process images on standard hardware. In the near future, it can be expected that general-purpose computers can handle volumetric images and/or image sequences with- out diffi culties. In the following, we will briefl y outline these key areas.

General-purpose computers now include suffi cient random access memory (RAM) to store multiple images. A 32-bit computer can ad- dress up to 4 GB of memory. This is suffi cient to handle complex image processing tasks even with large images. Emerging 64-bit computer sys- tems provide enough RAM even for demanding applications with image sequences and volumetric images.

While in the early days of personal computers hard disks had a ca- pacity of just 5–10 MB, nowadays disk systems with more than thousand times more storage capacity (10–60 GB) are standard. Thus, a large num- ber of images can be stored on a disk, which is an important requirement for scientifi c image processing. For permanent data storage and PC ex- change, the CD-ROM is playing an important role as a cheap and versatile storage media. One CDcan hold up to 600 MB of image data that can be read independent of the operating system on MS Windows, Macintosh, and UNIX platforms. Cheap CD-ROM writers allow anyone to produce CDs. Once cheap DVD+RW writers are on the market, a storage media with a even higher capacity of 4.7 GB, compatible to standard DVD (dig- ital video disks) ROM and video disks, will be available.

Within the short history of microprocessors and personal computers, computing power has increased tremendously. From 1978 to 2001 the clock rate has increased from 4.7 MHz to 1.6 GHz by a factor of 300. The speed of elementary operations such as fl oating-point addition and mul-

1.7 Components of an Image Processing System                             25

 

tiplication has increased even more because on modern CPUs these oper- ations have now a throughput of only a few clocks instead of about 100 on early processors. Thus, in less than 25 years, the speed of fl oating- point computations on a single microprocessor increased more than a factor of 10 000.

Image processing could benefi t from this development only partly. On modern 32-bit processors it became increasingly ineffi cient to trans- fer and process 8-bit and 16-bit image data. This changed only in 1997 with the integration of multimedia techniques into PCs and workstations. The basic idea of fast image data processing is very simple. It makes use of the 64-bit data paths in modern processors for quick transfer and processing of multiple image data in parallel. This approach to parallel computing is a form of the single instruction multiple data (SIMD) con- cept. In 64-bit machines, eight 8-bit, four 16-bit or two 32-bit data can be processed together.

Sun was the fi rst to integrate the SIMDconcept into a general-purpose computer architecture with the visual instruction set (VIS) on the Ultra- Sparc architecture [126]. In January 1997 Intel introduced the Multi- media Instruction Set Extension (MMX ) for the next generation of Pen- tium processors (P55C). The SIMDconcept was quickly adopted by other processor manufacturers. Motorola, for instance, developed the AltiVec instruction set. It has also become an integral part of new 64-bit architec- tures such as in IA-64 architecture from Intel and the x86-64 architecture from AMD.

Thus, it is evident that SIMD-processing of image data will become a standard part of future microprocessor architectures. More and more image processing tasks can be processing in real time on standard mi- croprocessors without the need for any expensive and awkward special hardware.

 

Software and Algorithms

The rapid progress of computer hardware may distract us from the im- portance of software and the mathematical foundation of the basic con- cepts for image processing. In the early days, image processing may have been characterized more as an “art” than as a science. It was like tapping in the dark, empirically searching for a solution. Once an algo- rithm worked for a certain task, you could be sure that it would not work with other images and you would not even know why. Fortunately, this is gradually changing. Image processing is about to mature to a well- developed science. The deeper understanding has also led to a more re- alistic assessment of today’s capabilities of image processing and analy- sis, which in many respects is still worlds away from the capability of human vision.

26                                                                                      1 Applications and Tools

 

It is a widespread misconception that a better mathematical founda- tion for image processing is of interest only to the theoreticians and has no real consequences for the applications. The contrary is true. The ad- vantages are tremendous. In the fi rst place, mathematical analysis allows a distinction between image processing problems that can and those that cannot be solved. This is already very helpful. Image processing algorithms become predictable and accurate, and in some cases optimal results are known. New mathematical methods often result in novel ap- proaches that can solve previously intractable problems or that are much faster or more accurately than previous approaches. Often the speed up that can be gained by a fast algorithm is considerable. In some cases it can reach up to several orders of magnitude. Thus fast algorithms make many image processing techniques applicable and reduce the hardware costs considerably.

 

1.8 Further Readings‡

 

In this section, we give some hints on further readings in image processing.

 

Elementary textbooks. “The Image Processing Handbook” by Russ [158] is an excellent elementary introduction to image processing with a wealth of application examples and illustrations. Another excellent elementary textbook is Nalwa [130]. He gives — as the title indicates — a guided tour of computer vision.

 

Advanced textbooks. Still worthwhile to read is the classical, now almost twenty year old textbook “Digital Picture Processing” from Rosenfeld and Kak [157]. Other classical, but now somewhat outdated textbooks include Gonzalez and Woods [55], Pratt [142], and Jain [86]. The textbook of van der Heijden

[188] discusses image-based measurements including parameter estimation and object recognition.

 

Collection of articles. An excellent overview of image processing with di- rect access to some key original articles is given by the following collections of articles: “Digital Image Processing” by Chelappa [19], “Readings in Computer Vision: Issues, Problems, Principles, and Paradigms” by Fischler and Firschein [41], and “Computer Vision: Principles and Advances and Applications” by Kas- turi and Jain [92, 93].

 

Handbooks. The “Practical Handbook on Image Processing for Scientifi c Ap- plications” by Jä hne [81] provides a task-oriented approach with many practical procedures and tips. A state-of-the-art survey of computer vision is given by the three-volume “Handbook of Computer Vision and Applications by Jä hne et al. [83]. Algorithms for image processing and computer vision are provided by Voss and Sü ß e [192], Pitas [139], Parker [135], Umbaugh [186], and Wilson and

Ritter [198].

1.8 Further Readings‡                                                                                              27

Textbooks covering special topics. Because of the cross-disciplinary na- ture of image processing (Section 1.5), image processing can be treated from quite diff erent points of view. A collection of monographs is listed here that focus on one or the other aspect of image processing:

Topic                                                                           References

Image sensors                                                            Holst [69], Howell [74],

Janesick [88]

MR imaging                                                                 Haacke et al. [59], Liang and Lauterbur [110]

Geometrical aspects of computer vision                   Faugeras [37]

Perception                                                                   Mallot [117], Wandell [193]

Machine vision                                                           Jain et al. [87], Demant et al. [27]

Robot vision                                                               Horn [73]

Signal processing                                                      Granlund and Knutsson [57], Lim [112]

Satellite imaging and remote sensing                       Richards and Jia [152],

Schott [165]

Industrial image processing                                      Demant et al. [27]

Object classifi cation and pattern recognition          Schü rmann [166], Bishop

[9]

High-level vision                                                        Ullman [185]

28                                                                                      1 Applications and Tools

 

 

Image Representation

Introduction

This chapter centers around the question of how to represent the infor- mation contained in images. Together with the next two chapters it lays the mathematical foundations for low-level image processing. Two key points are emphasized in this chapter.

First, the information contained in images can be represented in en- tirely diff erent ways. The most important are the spatial representation (Section 2.2) and wave number representation (Section 2.3). These repre- sentations just look at spatial data from diff erent points of view. Since the various representations are complete and equivalent, they can be converted into each other. The conversion between the spatial and wave number representation is the well-known Fourier transform. This trans- form is an example of a more general class of operations, the unitary transforms (Section 2.4).

Second, we discuss how these representations can be handled with digital computers. How are images represented by arrays of digital num- bers in an adequate way? How are these data handled effi ciently? Can fast algorithms be devised to convert one representation into another? A key example is the fast Fourier transform, discussed in Section 2.5.

 

⇐ Предыдущая234567891011Следующая ⇒

Поделиться: