PHYSICS-BASED INTEGRATION OF MULTIPLE SENSING MODALITIES FOR SCENE INTERPRETATION

The fusion of multiple imaging modalities offers many advantages over the analysis, separately, of the individual sensory modalities. In thi s paper we present a unique approach to the integrated analysis of dis parate sources of imagery Sor object recognition. The approach is base d on physics-based modeling of the image generation mechanism. Such mo dels make possible features that are physically meaningful and have an improved capacity to differentiate between multiple classes of object s. We illustrate the use of physics-based approach to develop multisen sory vision systems for different object recognition application domai ns. The paper discusses the integration of different suites of sensors , the integration of image-derived information with model-derived info rmation, and the physics-based simulation of multisensory imagery.