Evolutionary Design of Imaging AGC for ATR and Tracking

The US Army has identified a need to provide advanced AGC (automatic gain control) algorithms for ATR tasking involving a man-in-the-loop. To be practical, such provision must be achieved in such a way as to allow the continued on-demand production of high-performance RGC algorithms via a cost effective design approach. This proposal by AFT, Inc. addresses these concerns by providing a means to achieve significant advances in high performance AGC algorithm design for precisely the cases where the man is in the loop. In order to achieve this, AFT proposes to upgrade its highly successful existing system design package, STADIUM, to allow the evolutionary design of needed AGC algorithms by the Image Analyst or operator himself while avoiding the interminable problems of having to mathematically model the human visual system and other uncertainties in the image chain. AET, Inc. intends to accomplish this program objective by incorporating a design engine within STADIUM to allow evolutionary rather than mathematical design. The Image Analyst or operator will him or herself guide or evolve the on-going STADIUM AGC design process toward the most fit algorithm for the specific AGC tasking at hand by visually choosing the most fit alternatives at each instant of time. This research will lead to the commercialization of previously funded DoD and ARPA AGC and design automation research, and will focus these results on the production of techniques to allow continued development of automated AGC algorithms.