Exploring Multi scale Mathematical Morphology for Dark Image Enhancement

Dark image enhancement is an essential requirement in image processing domain. As images are captured in different illumination conditions, it is important to enhance dark images or color images suffering from lack of contrast. The present algorithmsfor contrast enhancement sometimes result in artifacts andunnecessary changes. Recently Rivera et al. proposed a content aware algorithm for dark image enhancement. The technique used by them is known as channel division. This paper proposes and implements a new technique which is then compared with the work of Rivera et al. The technique used to enhance gray scale images in this paper is “multiscale mathematical morphology”. This technique enhances the magnitude image constructed from RGB using huepreserving multi-scale morphological filters. By combining original direction cosine values and the enhanced magnitude image the enhanced image is obtained. A prototype application is built to test the efficiency of this technique. The results are compared with Channel Division technique. The experiments revealed that the proposed method outperforms the channel division approach. KeywordsContrast enhancement, multi scaling, mathematical morphology, local contrast enhancement ________________________________________________________*****_____________________________________________________