Reduction of computational costs for identification of the patch location on large-sized images

The use of the pseudo-gradient method for solving the problems of pattern search, identification and parameter estimation of image patch location is proposed. At that the reference and sought image patches may have relative spatial and brightness deformations. The structural optimization of search algorithms is used for computational costs reduction. The expressions for calculation of the image patch erroneous choice probability and the probability distribution of the offered algorithms iteration number are found.