9305630 Kimia This is the first year of a three-year continuing award. The research addresses the problem of shape description to support the recognition of objects from their two-dimensional profiles. Classically, object recognition is plagued with problems due to variations of the projected objects from the ideal model due to partial occlusion, movement of object parts, degradation due to noise, changes in viewpoint and viewing direction, shadows, and other effects. This project addresses these problems with a general purpose representation for shape which embeds these variations in the form of deformations and is derived from first principles. An object recognition system will be implemented based on this framework. The theoretical effort is focused on extending the framework to the analysis and representation of surfaces in three-dimensions. The final result should support object recognition for the complex problems that arise in industrial applications in addition to those that arise in more general machine vision.
