PERFORMANCE OF MINIMUM-MEAN-SQUARE-ERROR FILTERS FOR SPATIALLY NONOVERLAPPING TARGET AND INPUT-SCENE NOISE

Using computer simulations, we investigate the performance of a minimu m-mean-square-error filter for input-scene noise that is spatially non overlapping (disjoint) with a target for a limited set of images. Diff erent input-scene-noise statistics are used to test the filter perform ance. We show that in the presence of spatially nonoverlapping target and input-scene noise, the output of the minimum-mean-square-error fil ter has a well-defined correlation peak, small sidelobes, and a high p eak-to-correlation-energy ratio compared with other widely used filter s such as the classical matched filter, the phase-only filter, and the inverse filter. We also test the robustness of the minimum-mean-squar e-error filter to errors in noise statistics used in the filter design . We show that, for the images tested here, the performance of the min imum-mean-square-error filter is not sensitive to errors in noise stat istics and the filter can detect the target even if a considerable err or exists. The discrimination capability and the illumination sensitiv ity of the minimum-mean-square-error filter are also tested.