Real-Time Pattern Recognition Using Compressed Video

DESE proposes to adapt its previously developed Image Enhancement System (IES) to extract features and recognize images from high data rate image data streams. The IES has been developed for the Anti-Satellite (ASAT) program in order to provide robust and accurate aimpoints for ASAT missiles. The approach DESE proposes is to use the Landmark Feature Extraction (LFE) algorithm to produce a compact (compressed) representation of the items within a scene which can then quickly and efficiently be compared to a library of compressed templates to unambiguously identify the items. The LFE algorithm has been chosen because: (1) it has been shown to work with both visible and IR images, (2) it is able to compare large numbers of templates to the detected image, (3) it is invariant to rotation, translation, intensity, and scaling of the detected image, (4) it is focused on shape features, and (5) it runs on commercial computer and can be easily ported to other computer systems.