Spatial-Spectral Automatic Target Recognition

The complexity and pace of the modern battlefield make Automatic Target Recognition (ATR) technologies critical to the ability to fight and prevail. Given this, it has proven technically challenging to implement ATR in a universally effective and effective way. Shape-based ATR suffers many of the limitations of the human eye-brain system. For example, the algorithms can be deceived by decoys, camouflage, shadow or darkness. Systems that take advantage of spectral information may be resistant to these limitations while suffering from others such as mis-identification due to mixed pixels. This research effort seeks to redress the limitations of shape or spectral-based systems by synergistically combining the best features of both to create a superior ATR system. Photon Research Associates, while leading research in spectral ATR, recognizes that this program will require equivalent experience and innovation in shape-based ATR; therefore we have teamed with Hughes SCS in order to create a creative, effective team to address this problem. The initial effort will consist primarily of two activities: 1) devising a number of candidate spatial-spectral ATR algorithms which promise to provide performance superior to either method in isolation, and 2) testing these algorithms throughout the spectral, spatial and operational trade space.