A Bayesian Track-before-Detect Algorithm for IR Point Target Detection

An algorithm has been developed for the detection of point targets in uncluttered background based on a Bayesian track before detect method. The algorithm has an application in the detection of sea skimming antiship missiles at maximum range, when the missile appears over the horizon. Because of the long range, angular motion of the target will be insignificant, and target motion cannot be used to aid detection. The effect of filtering with a number of spatial filters on detection efficiency is assessed. The algorithm was tested on an infrared image sequence of an aircraft approaching the sensor at low level over water with a diffuse cloud background, and it was found to perform significantly better than simple detection by threshold exceedance. The algorithm is intended for application on a massively parallel processor where each pixel is assigned to a processing element, and each pixel is considered to be an individual sensor.