This fourth edition has been developed to reflect the changes that have occurred in techniques for the analysis of digital image data in remote sensing over the past five years or so. Its focus is on those procedures that seem now to have become part of the set of tools regularly used to perform thematic mapping. As with previous revisions, the fundamental material has been preserved in its original form because of its tutorial value; its style has been revised in places and it has been supplemented if newer aspects have emerged in the time since the third edition appeared. The theme of the book remains, however, on the needs of the senior student and practitioner.
The earlier editions have contained extensive material in Chapter 1 on satellite programs and sensor characteristics. Although that material is important in the context of understanding the application of image analysis procedures, the rapid development of quasi- and fully operational programs in the past decade has meant that expanding the material of Chap. 1, as required, would have distracted from the role of that chapter to introduce the reader to the nature and properties of digital image data in remote sensing. Accordingly, all of the material on satellite programs and
sensor specifications has been moved to a new Appendix A. Chapter 1 has then been completely rewritten as a stand-alone introduction to sensors in general and the data properties of importance in image analysis.
Many changes have been made throughout the book to meet the increasing emphasis on hyperspectral data and its analysis. Although much of that is contained in Chap. 13, techniques required for hyperspectral data processing are developed in the context of previous chapters, particularly new material on feature extraction tools that work well on hyperspectral data sets; they are covered in Chap. 10. Chapter 10 has been re-named. It was felt that there is too much confusion in the term Data Fusion to retain it as the title for material that is fundamentally concerned with thematic mapping from multiple data sources and multiple sensors. Chapter 8, dealing with supervised classification methods, has been substantially supplemented. Sections have been incorporated on k nearest neighbour classification, Markov random fields and support vector classifiers.
Other modifications relate to the noise adjusted principal components transformation, the definition of texture, a re-working of the contrast modification material and the inclusion of several further illustrations and problems.
The authors continue to enjoy the very strong support and understanding of their families, so important in the undertaking of this work, for which they express and record their sincere gratitude.