Dual-band ATR for forward-looking infrared images

Abstract

Three aspects of object recognition and tracking are evaluated on forward-looking infrared data from two wavebands. The combination of a correlator with a Kalman filter and a neural network embedded within the tracking loop is shown to increase true recognitions and decrease false recognitions. Large training sets may be reduced by condensing via a k-nearest-neighbour algorithm without significant loss of network performance. Systems combining information from the two wavebands show only a slight improvement over the best single band channel.