Read an Excerpt
The Essential Guide to Image Processing
Academic Press
Copyright © 2009 Elsevier Inc.
All right reserved.
ISBN: 978-0-08-092251-5
Chapter One
Introduction to Digital Image Processing
Alan C. Bovik The University of Texas at Austin
We are in the middle of an exciting period of time in the field of image processing. Indeed, scarcely a week passes where we do not hear an announcement of some new technological breakthrough in the areas of digital computation and telecommunication. Particularly exciting has been the participation of the general public in these developments, as affordable computers and the incredible explosion of the World Wide Web have brought a flood of instant information into a large and increasing percentage of homes and businesses. Indeed, the advent of broadband wireless devices is bringing these technologies into the pocket and purse. Most of this information is designed for visual consumption in the form of text, graphics, and pictures, or integrated multimedia presentations. Digital images are pictures that have been converted into a computer-readable binary format consisting of logical 0s and 1s. Usually, by an image we mean a still picture that does not change with time, whereas a video evolves with time and generally contains moving and/or changing objects. This Guide deals primarily with still images, while a second (companion) volume deals with moving images, or videos. Digital images are usually obtained by converting continuous signals into digital format, although "direct digital" systems are becoming more prevalent. Likewise, digital images are viewed using diverse display media, included digital printers, computer monitors, and digital projection devices. The frequency with which information is transmitted, stored, processed, and displayed in a digital visual format is increasing rapidly, and as such, the design of engineering methods for efficiently transmitting, maintaining, and even improving the visual integrity of this information is of heightened interest.
One aspect of image processing that makes it such an interesting topic of study is the amazing diversity of applications that make use of image processing or analysis techniques. Virtually every branch of science has subdisciplines that use recording devices or sensors to collect image data from the universe around us, as depicted in Fig. 1.1. This data is often multidimensional and can be arranged in a format that is suitable for human viewing. Viewable datasets like this can be regarded as images and processed using established techniques for image processing, even if the information has not been derived from visible light sources.
1.1 TYPES OF IMAGES
Another rich aspect of digital imaging is the diversity of image types that arise, and which can derive from nearly every type of radiation. Indeed, some of the most exciting developments in medical imaging have arisen from new sensors that record image data from previously little used sources of radiation, such as PET (positron emission tomography) and MRI (magnetic resonance imaging), or that sense radiation in new ways, as in CAT (computer-aided tomography), where X-ray data is collected from multiple angles to form a rich aggregate image.
There is an amazing availability of radiation to be sensed, recorded as images, and viewed, analyzed, transmitted, or stored. In our daily experience, we think of "what we see" as being "what is there," but in truth, our eyes record very little of the information that is available at any given moment. As with any sensor, the human eye has a limited bandwidth. The band of electromagnetic (EM) radiation that we are able to see, or "visible light," is quite small, as can be seen from the plot of the EM band in Fig. 1.2. Note that the horizontal axis is logarithmic! At any given moment, we see very little of the available radiation that is going on around us, although certainly enough to get around. From an evolutionary perspective, the band of EM wavelengths that the human eye perceives is perhaps optimal, since the volume of data is reduced and the data that is used is highly reliable and abundantly available (the sun emits strongly in the visible bands, and the earth's atmosphere is also largely transparent in the visible wavelengths). Nevertheless, radiation from other bands can be quite useful as we attempt to glean the fullest possible amount of information from the world around us. Indeed, certain branches of science sense and record images from nearly all of the EM spectrum, and use the information to give a better picture of physical reality. For example, astronomers are often identified according to the type of data that they specialize in, e.g., radio astronomers and X-ray astronomers. Non-EM radiation is also useful for imaging. Some good examples are the high-frequency sound waves (ultrasound) that are used to create images of the human body, and the low-frequency sound waves that are used by prospecting companies to create images of the earth's subsurface.
One commonality that can be made regarding nearly all images is that radiation is emitted from some source, then interacts with some material, then is sensed and ultimately transduced into an electrical signal which may then be digitized. The resulting images can then be used to extract information about the radiation source and/or about the objects with which the radiation interacts.
We may loosely classify images according to the way in which the interaction occurs, understanding that the division is sometimes unclear, and that images may be of multiple types. Figure 1.3 depicts these various image types.
Reflection images sense radiation that has been reflected from the surfaces of objects. The radiation itself may be ambient or artificial, and it may be from a localized source or from multiple or extended sources. Most of our daily experience of optical imaging through the eye is of reflection images. Common nonvisible light examples include radar images, sonar images, laser images, and some types of electron microscope images. The type of information that can be extracted from reflection images is primarily about object surfaces, viz., their shapes, texture, color, reflectivity, and so on.
Emission images are even simpler, since in this case the objects being imaged are self-luminous. Examples include thermal or infrared images, which are commonly encountered in medical, astronomical, and military applications; self-luminous visible light objects, such as light bulbs and stars; and MRI images, which sense particle emissions. In images of this type, the information to be had is often primarily internal to the object; the image may reveal how the object creates radiation and thence something of the internal structure of the object being imaged. However, it may also be external; for example, a thermal camera can be used in low-light situations to produce useful images of a scene containing warm objects, such as people.
Finally, absorption images yield information about the internal structure of objects. In this case, the radiation passes through objects and is partially absorbed or attenuated by the material composing them. The degree of absorption dictates the level of the sensed radiation in the recorded image. Examples include X-ray images, transmission microscopic images, and certain types of sonic images.
Of course, the above classification is informal, and a given image may contain objects, which interacted with radiation in different ways. More important is to realize that images come from many different radiation sources and objects, and that the purpose of imaging is usually to extract information about either the source and/or the objects, by sensing the reflected/transmitted radiation and examining the way in which it has interacted with the objects, which can reveal physical information about both source and objects.
Figure 1.4 depicts some representative examples of each of the above categories of images. Figures 1.4(a) and 1.4(b) depict reflection images arising in the visible light band and in the microwave band, respectively. The former is quite recognizable; the latter is a synthetic aperture radar image of DFW airport. Figures 1.4(c) and 1.4(d) are emission images and depict, respectively, a forward-looking infrared (FLIR) image and a visible light image of the globular star cluster Omega Centauri. Perhaps the reader can guess the type of object that is of interest in Fig. 1.4(c). The object in Fig. 1.4(d), which consists of over a million stars, is visible with the unaided eye at lower northern latitudes. Lastly, Figs. 1.4(e) and 1.4(f), which are absorption images, are of a digital (radiographic) mammogram and a conventional light micrograph, respectively.
1.2 SCALE OF IMAGES
Examining Fig. 1.4 reveals another image diversity: scale. In our daily experience, we ordinarily encounter and visualize objects that are within 3 or 4 orders of magnitude of 1 m. However, devices for image magnification and amplification have made it possible to extend the realm of "vision" into the cosmos, where it has become possible to image structures extending over as much as 1030 m, and into the microcosmos, where it has become possible to acquire images of objects as small as 10-10 m. Hence we are able to image from the grandest scale to the minutest scales, over a range of 40 orders of magnitude, and as we will find, the techniques of image and video processing are generally applicable to images taken at any of these scales.
Scale has another important interpretation, in the sense that any given image can contain objects that exist at scales different from other objects in the same image, or that even exist at multiple scales simultaneously. In fact, this is the rule rather than the exception. For example, in Fig. 1.4(a), at a small scale of observation, the image contains the bas-relief patterns cast onto the coins. At a slightly larger scale, strong circular structures arose. However, at a yet larger scale, the coins can be seen to be organized into a highly coherent spiral pattern. Similarly, examination of Fig. 1.4(d) at a small scale reveals small bright objects corresponding to stars; at a larger scale, it is found that the stars are non uniformly distributed over the image, with a tight cluster having a density that sharply increases toward the center of the image. This concept of multiscale is a powerful one, and is the basis for many of the algorithms that will be described in the chapters of this Guide.
1.3 DIMENSION OF IMAGES
An important feature of digital images and video is that they are multidimensional signals, meaning that they are functions of more than a single variable. In the classic study of digital signal processing, the signals are usually 1D functions of time. Images, however, are functions of two and perhaps three space dimensions, whereas digital video as a function includes a third (or fourth) time dimension as well. The dimension of a signal is the number of coordinates that are required to index a given point in the image, as depicted in Fig. 1.5. A consequence of this is that digital image processing, and especially digital video processing, is quite data-intensive, meaning that significant computational and storage resources are often required.
(Continues...)
Excerpted from The Essential Guide to Image Processing Copyright © 2009 by Elsevier Inc.. Excerpted by permission of Academic Press. All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
