![]() | Only 14 pages are availabe for public view |
Abstract Computer vision-based applications have attracted the attention of many researchers due to the frequent use of them in almost every field. These applications are designed for high-visibility input images. The images captured under low-light conditions suffer from low contrast, color vividness and low visibility of image details due to varying or insufficient illumination. Using these insufficient illumination images with the vision-based applications may degrade the performance of these applications. Therefore, it is necessary to enhance low-light images. Thus, several low-light image enhancement algorithms have been proposed in literature to deal with the problems of insufficient illumination. In this thesis, the enhancement of low-light images is considered based on Retinex model. The main contributions of the proposed work can be divided into two parts: First, an effective fusion-based low-light image enhancement algorithm is proposed based on the fusion of both maximum color and bright channels to overcome the color distortion and the halo artifacts problems. Secondly, an efficient structure-preserving low-light image enhancement (SPLIE) algorithm is proposed to estimate the optimal illumination map of the input image using a multi-objective optimization function. Experimental results demonstrated that the proposed algorithms improved the visibility of low-light images by enhancing the illumination while preserving the naturalness and structural details and avoiding over-enhancement. |