Ultra Lethal Targeting by Optical Recognition ULTOR

1The Navy desires to demonstrate an optical processing system capable of target aim point selection in an infrared image, in real time. Advanced Optical Systems, Inc. (AOS) has demonstrated with optical processing: aim point selection in visible imagery, in clutter; target recognition in infrared imagery; and small feature recognition internal to a target, in simulated IR imagery. We propose to combine lessons learned from each of these tasks with an innovative hybrid processing architecture, in order to provide robust aim-point selection in the infrared. The keys to successful real time aim-point selection in cluttered infrared images are: Adaptive control of image preprocessing; Proper design of filters; and Dual channel optical processing architecture. AOS proposes a correlator in which the preprocessor is controlled by the features of the Fourier transform plans and the output of the correlator. In this innovative approach a neural network will learn the relationship between preprocessor parameters and Fourier plane features, with quality aim-point correlation as the criteria. The result will be preprocessing which adapts so that the image presented to the correlator is modified to provide the best aim-point selection.