Terms and Conditions:

Robert Fisher wrote:This restoration of Ramakant Nevatia's Machine Perception textbook was funded by the EU's ECVision Network on Cognitive Computer Vision.

Book Excerpts:

The field of machine perception concerns the building of machines that sense and interpret their environments. This book is about visual perception. Potential applications for such systems include tasks such as automation of industrial processes of inspection and assembly, automated medical x-ray diagnosis, vehicle guidance and automatic photo-interpretation. Beginning from the analysis of simple polyhedral scenes in the early 1960s, the field has progressed to the point where useful analysis of complex natural and industrial scenes is possible and many practical prototype and commercial systems are available. Because of its immense potential applications, interest in this field has been growing rapidly.

Research literature in the field has multiplied but is scattered over many journals, conference proceedings, and research reports. While the active researchers seem to share much the same knowledge of previous work, a newcomer to the field has a difficult job in sorting out the vast literature. This book is aimed at easing this task by providing basic concepts, details of the major approaches, and a guide to the literature.

The field of visual perception is still maturing. Some aspects of the problems are fairly well understood and have a well developed theory which is described in detail in the text. For other problems, however, comprehensive techniques do not exist, and the literature consists of a large number of methods of limited utility. In such cases this book has grouped the techniques by their common themes and described the basic concepts. Varying amounts of detail are given for the specific techniques, based on my judgment of their generality and importance. This book has tried to provide fairly complete references. To simplify the logistics of acquiring the pictures, examples from author's own work or those of author'sstudents and colleagues have been provided, where applicable; similar examples could have been taken from others' work.

Intended Audience:

This book assumes no previous knowledge of the field and aims to provide a comprehensive knowledge of its methods. It is intended for use as a text for a one-semester graduate or senior-level course and also as a guide for the practicing professional.

This book does not discuss computer programs in detail, but one must be familiar with digital computer programming in order to fully appreciate the difficulties of mechanizing the described processes. The mathematical content of the text is small, and for an overview the mathematical parts can be skipped. However, for a detailed understanding of certain topics, knowledge of a variety of mathematical tools would be helpful. These tools include calculus, analytical geometry, matrix theory and linear algebra, numerical analysis and graph theory. Knowledge of freshman-level physics may also help.