Comparison of the maximum likelihood ratio test algorithm and linear filters for target location in binary images

We consider the problem of estimating the position of objects in binary ima ges corrupted with nonoverlapping and spatially nonhomogeneous noise. For t his application, we compare classical linear filters with the recently prop osed maximum likelihood ratio test (MLRT) algorithm, which is optimal in th e maximum likelihood sense for object detection. We first demonstrate that the MLRT algorithm can be approximated with a good precision by the square of a correlation product. We then compare the MLRT with the Classical Match ed, Phase-only and Optimal Tradeoff filters in terms of probability of corr ect location. We conclude that if they are properly regularized, linear fil ters can achieve a performance level comparable to that of the MLRT. This r esult is important in the design of optical correlators, which often implem ent linear filters with binary input spatial light modulators. (C) 1999 Els evier Science B.V. All rights reserved.