Image restoration a comparative study of some methods applied to color images

Abstract

The present work introduces five different methods to deal with digital image restoration. Particularly deconvolution by Richardson-Lucy method, Wiener filter, deconvolution with Gaussian priors in the frequency domain, spatial domain and the use of sparse priors. The Bayesian methodology is based on the prior knowledge of some information that allows an efficient modeling of the image acquisition process. The edge preservation of objects into the image while smoothing noise is necessary for an adequate model. Thus, we use five deconvolution methods to recover images, all of the presented images are contained on TID 2008, all of them were previously degraded by Gaussian noise and convolved with a disc point spread function (PSF) making reference to a typical fluorescence microscopy degradation. The principal objective when using restoration methods in the context of image processing is to eliminate those effects caused by the excessive smoothness on the reconstruction process of an image which is rich in contours or edges and also is important to consider the process time due to an improvement in this área could lead to a faster application. A comparison between the five methods is presented for a restoration process. This collection of implemented methods has been compared using different metrics such as SNR, PSNR, SSIM and process time. The obtained results showed a satisfactory performance and the effectiveness of the proposed methods on color space.