Feature measurement augmentation for a dynamic programming-based IR target detection algorithm in the naval environment

The detection of long range air targets in a Naval scenario using passive Imaging IR sensor is a task of primary importance for current and next generation Naval equipment. The authors have investigated Dynamic Programming based target detection systems utilizing the output of an image filter as the input to a likelihood classifier based on intensity alone. Variations of this technique have been proven to offer high sensitivity to dim targets though environmental characteristics in the Naval scenario can give rise to clutter induced false alarms. The work presented herein investigates augmentation of the intensity classifier with textural analysis techniques on IR imagery in the 3-5 micron waveband to assist in false alarm discrimination. It is shown that augmentation with a textural classifier can improve rejection of false alarms due to clutter. This work is apt of an ongoing program of IRST and Surveillance Sensor processing development.