Optimization of Soft Morphological Filters with Parallel Annealing-Genetic Strategy

As an important issue in signal processing field, filter design is essentially a multiple-parameter optimization problem. Because the searching process of pure simulated annealing is rather long, and pure genetic is easy to be premature convergent, combining the probabilistic jumping search ability of simulated annealing with genetic fast converge to some local minimum of the search space, this paper proposes an effective and easy-to-be implemented parallel annealing-genetic strategy for soft morphological filters design. According to the empirical results as well as comparison with conventional genetic and simulated annealing algorithms, the effective and global optimization ability of the proposed strategy are verified.