Blind Source Separation and Genetic Algorithm for Image Restoration

Abstract

Digital images often suffer from point spreading or blurring from both known and unknown filters or point spread functions. The sources of degradation can be lens point spreading, misfocus, motion, and scattering in case of x-ray images or atmospheric turbulence. Therefore a digital image can suffer blurring from a single or a combination of various point spread functions, for example many images suffer from lens out of focus blur because of manufacturing limitations or satellite/aerial images suffer from lens focus and atmospheric turbulence etc. The obvious requirement of an imaging system is to reproduce an image that is as close to original as possible. Most existing image restoration methods uses blind deconvolution and deblurring methods that require good knowledge about both the signal and the filter and the performance depends on the amount of prior information regarding the blurring function and the signal. Often an iterative procedure is required for estimating the blurring function such as Richardson-Lucy method and is computational complex and expensive and sometime unstable. This paper presents a blind image restoration method based on techniques of blind signal separation (BSS) in combination with the genetic algorithm for parameters optimization. The method is not only simple but also requires little priori knowledge regarding the signal and the blurring function