The Use of Adaptive Sample Sizes of Objective Functions in the Algorithms of Pseudogradient Estimation of Interframe Spatial Distortions of Images

摘要

Algorithms for pseudogradient estimation of parameters of interframe spatial distortions of images, where local samples of the objective functions have adaptive sizes, are suggested and studied. It is shown that such algorithms make it possible to substantially reduce computational burden in comparison with the algorithms with constant sample size providing the same accuracy of estimation. Particular constructions of the algorithms and examples of their application are given.