Perceptual Organization and Visual Recognition

COMPUTER VISION is a field of research that encompasses many objectives. A primary goal has been to construct visual sensors that can provide general-purpose robots with the same information about their surroundings as we receive from our own visual senses. This book takes an important step towards this goal by describing a working computer vision system named SCERPO. This system can recognize known three-dimensional objects in ordinary black-and-white images taken from unknown viewpoints, even when parts of the object are undetectable or hidden from view. A second major goal of computer vision re­ search is to provide a computational understanding of human vision. The research presented in this book has many implica­ tions for our understanding of human vision, particularly in the areas of perceptual organization and knowledge-based recogni­ tion. An attempt has been made to relate each computational result to the relevant areas in the psychology of vision. Since the material is meant to be accessible to a wide range of inter­ disciplinary readers, the book is written in plain language and attempts to explain most concepts from the starting position of the non-specialist. vii viii PREFACE One of the most important conclusions ansmg from this research is that visual recognition can commonly be achieved directly from the two-dimensional image without any prelim­ inary reconstruction of depth information or surface orienta­ tion from the visual input.

1. Introduction.- 1.0 Overview of spatial vision.- 1.1 Two viewpoints on computer vision.- 1.2 A demonstration of perceptual organization.- 1.3 Specific functions of perceptual organization.- 2. Previous Research.- 2.1 Gestalt psychology and perceptual organization.- 2.2 The principle of simplicity.- 2.3 Grouping as the formation of causal relations.- 2.4 The role of grouping in computer vision systems.- 3. The Significance of Image Relations.- 3.1 Probability of accidental occurrence.- 3.2 Limiting computational complexity.- 3.3 Conclusions.- 4. The Segmentation of Image Curves.- 4.1 Previous research on curve segmentation.- 4.2 Significance of a curve segmentation.- 4.3 Selecting the most significant structures.- 4.4 Demonstration of the algorithm.- 4.5 Evaluation and future research.- 5. The Use of Viewpoint Invariance.- 5.1 Three-space inferences from image features.- 5.2 Recovery of 3D properties from line drawings.- 5.3 A demonstration of three-space inference.- 5.4 Conclusions and future development.- 6. Model-based Search and Inference.- 6.1 Searching the space of possible viewpoints.- 6.2 Searching the space of possible objects.- 6.3 Summary.- 7. The Verification of Interpretations.- 7.1 Viewpoint determination in human vision.- 7.2 Definition of the problem.- 7.3 Previous research on viewpoint determination.- 7.4 Formulation of perspective projection.- 7.5 Newton-Raphson convergence.- 7.6 Solving for model parameters.- 7.7 Matching lines instead of points.- 7.8 Implementation and future research.- 8. The Scerpo Vision System.- 8.1 Edge detection.- 8.2 Perceptual organization.- 8.3 Matching and evidential reasoning.- 8.4 Verification.- 8.5 System performance and future extensions.- 9. Conclusions.- 9.1 Directions for future development.- Bibliographic Index.