GEOMETRIC MODELING AND COMPUTER VISION.

The central theme of the discussion is geometric matching for general purpose vision systems. Computer vision and geometric modeling are briefly introduced in broad, general terms to explain why geometric matching is needed and why geometric modeling has not addressed this problem. General geometric representation principles are introduced next in terms of common geometric forms and geometric primitive composition methods followed by a general definition of the geometric matching problem facing model-based computer vision systems. Subsequent sections investigate geometric matching issues in increasing spatial dimensions from points to curves to surfaces to volumes. Individual geometric primitive representations from the geometric modeling literature and any special matching considerations are treated for each dimension in the appropriate section. Other miscellaneous shape representations from the vision literature and texture and fractal representations are also discussed. Conclusions and unresolved problems are summarized in the final section.