Traditional LP algorithm based-on stack processing lies in two disadvantages for finding optimal solution of nonlinear filters: (1) Joint statistics of the signal and noise processes must be known; (2) The complexity of the LP increases faster than exponentially as a function of the window size. Base-on optimal stack filtering theory

the multi-stage adaptive WOS filter is proposed by expressing the PBF of WOS filter in term of linear separate function. Above disadvantages can be overcome by this algorithm

therefore

the class of filter is suitable to image processing.