A modified simple genetic algorithm for evolutionary image filters

Evolutionary algorithms are widely used optimisation technique for an efficient combinatorial design in the field of evolvable hardware. This paper deals with the design of non-linear image filter using genetic Algorithm. First, a modified simple genetic algorithm is proposed which is based on the concept of generation runs and shuffling operation. This generation run polls for the best fit offspring to pass on to the next generation along with retaining the elite population. This enables the Genetic algorithm to search the sub spaces of the large solution space. The shuffling operation at the end of each generation will avoid the problem of positional locality reference which causes the local minimum problem. A sample test function is employed to guarantee the convergence of sGA to an optimum solution. Second, the proposed modified sGA is capable of designing an efficient image filter for the impulse noise filtering. The suitable impulse noise modelling that reduces the computational complexity is also employed in the filter design.