OPTIMAL DETECTION OF A TARGET WITH RANDOM GRAY LEVELS ON A SPATIALLY DISJOINT BACKGROUND-NOISE

We describe a pattern-recognition processor that is optimal for detect ion and location of a target with white Gaussian random gray levels on a white random spatially disjoint background, me show that this algor ithm consists of correlations of the silhouette of the reference objec t with preprocessed versions of the scene image. This result can provi de a theoretical basis for pattern-recognition techniques that use non linear preprocessing of images before correlation. (C) 1996 Optical So ciety of America