Parallel Processing Benchmarks for Howard Cascading Computer Systems Phase II

The goal of this Phase 2 SBIR proposal is to develop an innovative LIDAR and image processing data fusion solution which addresses current government and commercial GIS and image processing needs. The proposed solution makes use of Massively Parallel Technology''s high performance parallel processing architecture in conjunction with state of the art mathmatics such as Zernike moments, Delaunay triangulation, foliage recognition algorithms, Hough transformations, Histogram equalization, bare-earth generation and break-line detection algorithms integrated with previously shown target recognition algorithms. The proposed work will be done using DISP (or similar) images with LIDAR images of the same geographic area to perform automated orthorectification, automated target detection (with slant angle correction), automated change detection, and multiple frequency band data fusion. This effort will further show the advantages of being able process multiple gigabyte dataset in RAM without disk swapping.