Arman Yaraee Undergraduate Dissertation 2014/15

Image deblurring

Supervised by J.Winkler

Abstract

Although there has been numerous developments in image deblurring, there is a considerable lack of performance. Deblurring is an inverse problem and by nature, inverse problems are highly unstable, therefore each image will perform differently. Image deblurring is a useful process before further processing an image. Distorted images will not perform as well under image interrogation techniques hence deblurring is one of the primal image processing tasks. Most of the techniques introduced in the past rely on the availability of the point spread function (PSF). This function describes the way blur has been added to the image. In the case of an unknown PSF, it is extremely difficult to reverse the process. Blind deconvolution introduces a technique which attempts to estimate this function. This project is aiming to inspect various image deblurring techniques and compare the result and performance of such methods. In addition these algorithms will be tested against a newly developed algorithm by Joab Winkler to determine the strongest deblurring method