Regionally equalized and contextually clipped gamma correction approach for dark image enhancement

In this paper, an efficient statistical approach, employing a highly adaptive gamma correction based on regionally distributed and independently equalized histograms for all regions followed by contextual clipping, is presented for overall enhancement of low contrast dark images keeping their intrinsic features preserved. For this purpose, input image is uniformly subdivided into several non-overlapping equal sized regions. A good estimation has been achieved by classifying these regions into three groups namely corner regions, boarder regions as well as inner regions. Further, separate histogram equalization can be performed, followed by individual region's contextual clipping so that unwanted domination of high frequency bins over other bins can be avoided. Later on, a non-linear transformational mapping has been imposed by suitable gamma-correction using required gamma value-set, which itself is derived by cumulative distribution of the intensity values in adaptively equalized histogram. The proposed methodology clearly outperforms other state-of-the-art methods in terms of complexity as well as quantitative and qualitative performance; and hence, can be appreciably used for a wide and dynamic range of image-database which belongs to various domains ranging from biomedical images to remotely sensed satellite images.