Neuro-fuzzy system based on particle swarm optimization algorithm for image denoising application

In this paper, we investigate the Neuro-Fuzzy System (NFS) design based on Particle Swarm Optimization (PSO) algorithm. The problem being studied concerns the optimal estimation of structure and parameters network. The common training algorithms such as gradient descent techniques are frequently used for NFS. However, they cannot possibly find the global optimum, which declines the network performance. The PSO is an optimization tool favoring global search in the feature space, constitutes therefore a more suitable method. The main purpose is to use the outstanding features of PSO in NFS training for any image processing function approximation. As illustration, we consider image denoising. The performance of the proposed method is validated on an image set and a comparison with other techniques is done. Experimental results prove the effectiveness of our approach and demonstrate that such system is strongly adaptive with respect to the noise type and leading to good restored images.