Characterization of the performance of the track-before-detect approach to moving target detection

In the pursuit of detecting ever smaller and dimmer targets, the track-before-detect methodology has been used to integrate the target energy through a time sequence of frames. However, when target tracking is required, an additional uncertainty over detection of a stationary target is introduced. This loss is characterized by the use of a detector optimum for Gaussian white noise for point targets constrained only by its maximum velocity (V). The authors characterize the gains and losses associated with the input target SNR, the number of frames integrated, and the uncertainty introduced by target movement constrained by a maximum velocity V, by comparison of the ratio of the SNR with respect to the SNR of single frame detection and known-target-path detection techniques. >