Search

In this chapter, we will summarize some technologies about image processing related to biometrics. Digital image processing will be simply introduced in Section 3.1. Section 3.2 discusses some kinds of image transformation. Section 3.3 contains the concept and methods of image enhancement. Image restoration and data compression are introduced in Section 3.4 and Section 3.5, respectively.

The second part of this book describes the design of high-voltage integrated circuits. One of the most common solutions for high-voltage applications using the SVX devices, involves complex low-voltage signal processing communicating with an external high-voltage environment. This communication is usually performed by high-voltage digital or analog output interfaces. These are discussed in this chapter, where circuit design techniques overcoming the voltage limitations of the devices are presented as well.

When the capabilities of present-day artificial information processing systems are compared to the capabilities of biological systems a few striking conclusions have to be drawn. Our present-day computers are mainly built with serial processors, which can process only one instruction on one data element at a time but they achieve a very high clock speed and accuracy. In this way they can perform a whole set of data-manipulation operations on very large data sets at very high speed. In order to solve a given problem an explicit algorithm or solution strategy has to be conceived, that when correctly programmed, yields a correct result after execution. In administrative applications, in scientific simulations, and for certain signal processing applications this computing power has been applied very successfully and has supported much of the technological and scientific progress of the last decennia. But in several application fields, however, the impact of the introduction of computers has not led to the expected solution of problems.

Fueled by the power and flexibility of digital signal processing techniques, new video systems in both consumer and industrial applications are increasingly making use of high-performance data conversion integrated circuits (ICs). Video-rate analog-to-digital and digital-to-analog converters commonly feature 10-bit resolutions and sampling rates upwards of 18 MHz. Progress in integrated antialiasing filters for such applications has not, however, kept pace. As the demand for low-cost, yet high-performance, video systems grows, such filters could well present a system bottleneck, both in performance and price. Perhaps the most difficult challenge lies in the area of High-Definition Television (HDTV) where superior video image quality is desired while maintaining, as far as practicable, near conventional transmission channel bandwidths. Japan Broadcasting Corporation’s MUSE system, for example, requires antialiasing filters for baseband signals with 8.15 MHz bandwidth and 10-bit linearity [1].

In EBT, any attempt to measure IC waveforms correctly using the SEM is hindered by two separate processes, namely; (1) signal distortion due to the finite spatial size of the electron beam and its temporal duration when pulsed and (2) the addition of noise. In this chapter, these processes are analyzed and an examination of techniques to counteract them is presented. Generating IC images by operating the electron beam tester in scanning mode has important application to locating and registering points of interest as well as detecting voltage and timing faults. The hardware and algorithms for processing these images are discussed.

The term “basic techniques” is used to include all the point operations which can be carried out on an analogue or digital image.

Data structures, and aspects of programming languages and programming systems that are relevant to image processing of astronomy data are discussed. Since parallel processing promises to greatly increase the speed with which images can be analyzed, techniques that will help achieve this speed and new methods that may become possible because of it are of interest.

New discoveries in the life sciences are often linked to the development of unique optical tools that allow experimental material to be examined in new ways. We as microscopists are constantly searching for new techniques for extracting even more optical information from the material we work with, as the subject of this book aptly demonstrates. It is not surprising then that microscopists have begun to turn to computer technology in order to squeeze more information from their experimental images. Computer processing can be used to obtain numerical information from the microscope image that is more accurate, less time-consuming, and more reproducible than the same operations performed by other methods. Computer processing can be used to enhance the appearance of the microscope image, for example to increase contrast or to reduce noise, in ways that are difficult to duplicate using photographic or video techniques alone. When used to their fullest power, digital image-processing techniques can be used to produce information about the optical image that cannot be obtained in any other way.

Image processing is usually approached as the problem of applying numerical algorithms to digitized gray-scale images and may be very expensive in terms of memory space and computation time. Therefore, it is worthwhile to look for more economical solutions. An analysis of typical image-processing tasks reveals that the usual costly approach can often be avoided or, in certain cases, may not even be appropriate. These considerations led us to the design and development of a special “Picture Processing Display Terminal” (PPDT) which has been devised for performing most of the required pre-processing and feature-extraction steps at the periphery of the computer at a small fraction of the costs that arise when the same processing is carried out in the computer. Hence the input to the computer already consists of binary representations of significant features, thus considerably reducing the amount of data to be further processed. Furthermore, each preprocessing step, which may be performed either with the man in the loop or automatically, is made visible in order to provide the feedback and the opportunity to evaluate the performance of the system and correct mistakes. The PPDT is part of a system for “distributed image processing”. The considerations leading to the design of such a system are discussed in this paper, as well as its possibilities and limitations.

We present a brief review of some new methodologies in digital image processing. The aim of image processing is to “understand” automatically images and 3-D scenes by computer. The various steps leading to this point are commonly divided in two major parts: image processing, properly speaking, and image analysis. Image processing techniques have been already largely developed and are well known from the astronomers’ community so that we will emphasize image analysis techniques and simply mention some new results in image processing which may be- of interest to the reader.

Fast digital algorithms are required in many applications of digital image processing, especially when many images of large si- ze are involved (as in astronomy data analysis) or real-time im- plementation is needed.

Parallel processing is an ill-defined concept. To some, parallelism implies little more than duplication of arithmetic units, or some sort of provision for simultaneous operation of various parts of an otherwise conventional computer architecture. For others, parallelism is represented by ‘pipelining’, in which the data streams through successive computational units, each carrying out its own particular operation on the data. Again, a parallel system can be one in which the basic machine instructions involve simultaneous access to a substantial subset of the total data to be processed.

This paper discusses some aspects of image processing — first with a high level look at image processing, its inputs, outputs and operations. Then, a specific research project is examined to illustrate how various image processing functions are used to derive results from image data. The project is the construction of a catalog of galaxies from photographic plates. The functions illustrated include correcting for the nonlinear plate response, modelling and removing the background with a polynomial, removing bright objects, histograms, smoothing, object detection, object parameter estimation, and manipulation of tabular data.

Many data in biomedicine are presented in the form of images. To name a few, there are radioisotope (RI) images, x-ray images, thermograms, microscope images, and ultrasound images. The processing of these images has been done mostly by human observers. The number of images to be processed is rapidly increasing due to the extension of health care to preventive and environmental medicine. Since the number of doctors and technicians is limited, there is a pressing need for automatic processing of these images. Although there have been a few elegant analog image processors, analog image processing has been limited to very specialized fields. Digital processing is more flexible, adaptable, accurate and reproducible. The development of integrated circuits, especially large scale integration, has steadily reduced the cost of digital processing even during the time of world-wide inflation. Hence digital image processing in biomedicine will be an indispensable part of any automated system.

We first review briefly the four major areas of image processing: computer graphics, efficient coding, restoration and enhancement, and pattern recognition. Then we discuss some general trends in digital image processing research.

In many ways, the light microscope is a highly perfected instrument. It produces an image resolved to a fraction of the wavelength of light, with excellent contrast, and with aberrations reduced to a bare minimum. It portrays images in many contrast modes and allows measurements of various optical quantities nondestructively. Still, practitioners of modern microscopy are continually striving to register fainter images, explore finer details, and record subtler contrast. We strive to detect weaker fluorescence, measure lower birefringence, and reveal miniscule absorbances and optical path differences, all with an effort to uncover dynamic physical and chemical details in finer and finer domains of the specimen.