Algorithm for adaptive pyramidal filtering of images

An algorithm for adaptive image filtering with the use of a pyramidal repre sentation of the initial image is presented. Each level of the pyramid can be adopted without change or may be subjected to processing, depending on h ow noisy it is. A method of processing the levels is chosen with considerat ion of the type of environment - boundary, near-boundary, and homogeneous. The idea of adaptation extends to all levels of the pyramid, making this al gorithm suitable for suppression of both high- and low-frequency noise and distortion. The results of an experimental study of the quality of filterin g of test images by the proposed algorithm and its classical adaptive analo gs are presented. Adaptive pyramidal filtering enjoys an advantage over the best classical algorithms and, unlike them, it does not require sensitive iterations in order to achieve the best result. (C) 2000 The Optical Societ y of America. [S1070-9762(00)00410-3].