A two-fold fusion fuzzy framework to restore non-uniform illuminated blurred image

Abstract

Despite significant progress in the field of imagery, the image captured may experience some deterioration that is mainly summarized in non-uniform illumination and blur. This paper proposes a new two-fold fusion fuzzy framework to restore non-uniform illuminated blurred image: Fuzzy Rule-Based Method to enhance Illumination (FRMI) and Fuzzy Rule-Based Method for Deblurring (FRMD). The purpose of the restoration is to approximate an image as similar as possible to the original image from the observed object. In the first stage (FRMI), a fuzzy rule-based methodology is proposed to sharpen the non-uniform illuminated image which is based on the local brightness differences among pixels by distinguishing the dark and bright image regions. To deblur the image, blind deconvolution method is used. The goal of the second stage (FRMD) is to use the proposed edge detection and fuzzy model to find the correct size of the point spread function (PSF) which is needed for the blind deconvolution method. Implementation of the proposed method on two different datasets consists of non-uniform illuminated blur image shows that the proposed method outperforms the other enhancement techniques. To measure the restored image quality, seven image quality measurement metrics are used. The proposed fusion method produces an even illuminated deblurred image with improvement in the details and contrasts.