Optimal morphological filter design using a bacterial swarming algorithm

Noise removal is an underlying issue of image processing. This paper proposes a generic approach to design an optimal filter which combines linear and morphological filtering techniques, so that both Gaussian and non-Gaussian noise can be rejected. The optimisation process is performed by a bacterial swarming algorithm (BSA), which is derived from the bacterial foraging algorithm (BFA) and involves underlying mechanisms of bacterial chemotaxis and quorum sensing. The performance of the combined filter optimised by BSA is analysed in comparison with the filter optimised by the genetic algorithm (GA), as well as with other commonly used filters. The simulation results demonstrated in this paper have shown the merits of the proposed filtering technique and the optimisation algorithm.