Particle swarm optimization based regularization for image restoration

Image restoration from a degraded observation has been a long standing problem in image processing. It requires a direct inversion of the degradation function in frequency domain which is ill posed in nature. So regularization has been used in the restoration process. The selection of regularization parameter still remains a difficult problem due to the amplification of noise in the inversion process. In this paper, we have proposed a PSO based regularization technique which adapts the regularization parameters depending on the noise and blurring conditions in the degraded image. Experimental results are presented to validate the efficiency of the proposed scheme.