Optimizing Stack Filters Based On Mirrored Threshold Decompositions

Stack filters are a class of sliding window, nonlinear digital filters. This class filters have two main popularity are threshold decomposition and stacking property. On the base of traditional stack filters, mirrored threshold decomposition is proposed. The research of stack filter makes it possible to be re-understanding of many of the traditional nonlinear filter from another angle. Its virtues lie in: applying the architecture of threshold decomposition, suitable to parallel processing and VLSI realization. It is including many kinds of nonlinear filters, being an important tool for nonlinear research. This is why stack filters have become the most representative and the very broad development filters in the field of non-linear filters.We have got a lot of results of the traditional stack filter optimization, some results of optimization algorithm is ideal. This paper mainly focuses on the optimization design of stack filters based on the mirrored threshold decomposition, using the new Global Optimization Algorithm. With different filtering criteria and optimization of structural optimize stack filter was studied. The main contents are as follows:The fundamental theory and realize processing of stack filters based on the mirrored is summarized in this dissertation. The optimization model based on MAE criteria is presented in this paper. PSO algorithm is used to optimize optimal stack filters. Compare with the traditional threshold decomposition stack filter, The former shows that the optimal stack filters can suppress noise and protect the details of image effectively, which improve filtering capability.But the optimization design of stack filters based on the mirrored threshold decomposition is more difficult then traditional stack filter. In order to solve this problem use Clone Selection Algorithm and PSO algorithm combined with the cloning algorithm, and made a series of progressive base on stack filters. In this dissertation I bring adoptive arithmetic operators, and bring forward second step search which can jump out the local best result automatically. The algorithm converges quickly, and has satisfactory capabilities of global and local search. The results of experiments show that the optimal stack filters can suppress noise and protect the details of image effectively, which improve filtering capability.According to the difference between the traditional stack filter and mirrored stack filter, A fast algorithm is proposed, which Reduce unnecessary units of positive Boolean function (PBF) and the correspond cost function units. This reduced the search time and enhanced the effectiveness of the algorithm.In the final chapter in order to further enhance the mirrored stack filter filtering capability. The optimization model based on MSE criteria is used in this paper. Using improved Clone Selection Algorithm; Simulation results show that on the base of MSE criteria, their ability to suppress noise is improved.