Improvement in PSF Estimation Accuracy in Blind Image Restoration using PSO

An image can convey lots of information about the picture quality. Picture quality is the most notable and distinctive feature of the image providing an information in the specified area. An image can also be identified by its pixel range, shape, of an image but level information retrieval is very high in case of the restoration. Image restoration is a key aspect related to image processing generally categorized as verification and identification or (recognition). Motion blur, out of focus, camera shake are some common distractions to human appearance becoming the pitfall for the image restoration system. For the realization of image restoration systems, one of the well-known feature extraction methods is Blur Stein’s Unbiased Risk Estimation (SURE). This method is based on Point Spread Function (PSF) estimation. This method is a filtered version to calculate the PSF from the degraded image. Blur SURE method is usually envisaged as wiener filtering process. It minimized the blur MSE and then applies blur-SURE estimation over blur MSE. To enhance result for point spread function here particle swarm optimization technique is used. For the estimation of the PSF, particle positions are calculated by the PSO. Parametric form of the gaussian kernel is used for the estimation of function. When these parameters are calculated then PSF estimation is done. SURE LET deconvolution is used for the whole process with particle swarm optimization is done for deconvolution process of blind and non blind.