Evaluations on Different Window Size towards the Performance of Block Overlapped Histogram Equalization Method

Abstract

Local enhancement methods are mostly needed for enhancing the contrast of relatively small objects in the image, or image with poor illumination. In this paper, we investigate the performance of one well-known local enhancement method, which is Block Overlapped Histogram Equalization (BOHE). In order to execute BOHE, one parameter, which is its window size, must be set. Therefore, in this paper, we investigate the effect of window size towards the performance of BOHE. This evaluation is based on five image quality measures, which are the entropy value, maximum distance, speckle index value, mean structure similarity index map value, and a new image quality measure proposed by Song-Der Chen in 2012. It is found that bigger window size produces more natural images as compared to the results from smaller window size.