4h-Sic GTO Device Structural With Improved Reliability at High Blocking Voltage

Not Available Existing approaches to the detection of vehicles from SAR imagery are handicapped by a high false alarm rate, especially when used in high clutter environments such as urban areas and forested regions. In this Phase I proposal. Image Corp, Inc. proposes a system for SAR-based vehicle detection that makes extensive use of contextual information derived from the SAR image itself. Multi-resolution and fast non-Gaussian algorithms will be used to make the processing both robust and fast. The key to achieving robustness is the construction and use of a detailed 2-D site model from the SAR image, which delineates roads, buildings, trees, shadow regions and open terrain. Candidate vehicles are detected using a Constant False Alarm Rate (CFAR) technique followed by filtering and post-processing. False alarms resulting from man-made structures such as buildings as well as those resulting from vegetation are detected using the site model. Image Corp's proposal is based on the work at the University of Maryland of the research group of Prof. Rama Chellappa, whose outstanding contributions to SAR image exploitation are widely recognized. The proposed work will be invaluable in improving battlefield awareness and in peace-keeping efforts. Potential commercial applications include transportation monitoring and disaster relief.