USC IMAGE UNDERSTANDING RESEARCH: 1994-1995

摘要

This paper summarizes the USC Image Understanding research projects and provides references to more detailed descriptions of each one. We have undertaken a broad range of research with a major effort directed toward developing image understanding techniques to extract and use high-level descriptions from images. These techniques allow us to infer three-dimensional shape descriptions from one or more intensity images, or from range data. The resulting descriptions are used for recognition and pose estimation. In addition, we briefly report our progress on building detection and change analysis supporting the RADIUS program, incorporation of knowledge representation techniques in computer vision programs, and an effort in implementing vision algorithms on parallel systems.