Fusion Algorithm Creation for Advanced Detection via Evolution (FACADE)

Sensor technology has advanced to the point where it is has become feasible to use multiple sensor modalities on the same platform. However, methods for intelligently fusing the data from the disparate sensors into a more descriptive form have proven elusive. We have developed a set of features, applicable to multiple sensor modalities, which are candidates for sensor fusion. These include blob features, Gabor features, spin image features, corners, and vertical surfaces. Fusion’s goal is to produce more accurate and informative registration, target detection, and identification. Our approach performs the fusion and classification for these sub-tasks using genetically inspired algorithms, producing robust multimodal registration, target detection and target identification algorithms. The resulting algorithms will provide insight into the features necessary for robust ATR, as well as a strategy for fusing the features and building a classifier from them. We also provide an efficient method of generating truth data for registration, detection and identification, and provide an evaluation plan for assessing the accuracy of the developed algorithms. Finally, by examining ATR accuracy as a function of the inclusion and exclusion of various sensing modalities, we can determine the best sensor configurations and fusion operations for performing ATR.