The configuration of stack filters by probabilistic search

An efficient search method for the configuration of stack filters is presented. Nonlinear digital filters such as these are known to be flexible and effective in applications where it is difficult to characterize the noise type. An open problem associated with designing stack filters is finding the right configuration for a given noise type and the signal characteristics which need to be preserved. The problem of finding the right configuration is modeled as a combinatorial search problem. An efficient search method is developed which considers the probability distribution of the objective function values. Based on this, the number of steps needed to come up with a good configuration is established. The required number of steps can be carried out in parallel in any type of parallel computer. The method is empirically verified, and the performance of the stack filters found by the search method is compared to the performance of the median filter