Multiframe subpixel target detection in the presence of fluctuating background and noise

This paper presents a performance of the known matched subspace detector (MSD) for video-based detection systems. It is well known that a subpixel target occupies only a part of the pixel area. Then there is a difference between the variances under two statistical hypotheses H0 and H1. This fact is a cause of the detector performance degradation. We analyze two opposite phenomena related with an increasing size of the target. First, the target's energy is increased, and, second, the pixel background variance is decreased. The performance degradation has been evaluated when there are the mismatches between the actual and designed parameters. It has been shown that the performance depends strongly on the subpixel background fill factor. These results have been obtained analytically and have been also justified via Monte Carlo simulations for an example scenario.