High Efficiency, Scalable, Parallel Processing Approaches for Multi-Sensor Data Fusion

"As the number and power of imaging resources grow there is an increasing need to automate the analysis of those images. With the advent of multi, hyper, and ultra-spectral image data the need to fuse this data into an automated single composite view foranalysis purposes also increases. Massively Parallel Technology, Inc. has a unique COTS computer clustering technology that has already shown 100x or better performance (compared to single workstation technologies) for a wide variety of image processingalgorithms using only 127 processors. This technology will be used to prove that sensor fusion algorithms can show similar speed increases. 1) Define a new hardware/software solution to provide object identification from a mixture of sensors using COTSequipment.2) Show a 100x or better performance gain in object identification over current workstation solutions.3) Show that this arbitrary rotation angle object identification is non-intrusive to existing systems, as well as lending itself to near real-time data processing.4) Show an automated figure of merit generation for each object found5) Show that tactically important information can be automatically extracted from an image."