Real-Time, Super Resolution Pixon Processing for MDA Kill Vehicles

We propose to implement image-processing algorithms that greatly enhance resolution of closely spaced objects (CSOs), sensitivity, and tracking accuracy in real-time missile-guidance systems. These enhancements are accomplished without prior knowledge of the number and composition of the targets within a CSO cluster. The proposed capabilities are relevant to the MDA EKV program and complementary to competing approaches for separating CSO targets. The algorithms are based upon our proprietary Quick Pixon method, already implemented in hardware and demonstrated to increase resolution by up 2X and decrease background noise by 6X. It will also include DigiVision’s spatially adaptive contrast enhancement that optimally prepares images for the dynamic range of the human eye and machine vision systems. The program merges the capabilities proposed in two independent Phase 1 programs (DASG60-02-P-0139 and W9113M-04-P-0109), and will provide 16-bit accuracy for IR sensors, 10-bit accuracy for full color EO sensors, and large format capability (up to 150 Megapixels/sec). By the conclusion of Phase 2, all of these capabilities will be implemented in a single Field Programmable Gate Array design that is tested and proven in preparation for moving this design to an Application Specific Integrated Circuit in Phase 3.