Digital Image Enhancement Through Bi-Histogram Equalization Using Entropy-Based Plateau Limit

The equalization of histograms is a simple and efficient process for contrast enhancement. This paper presents equalization of the bi-histogram with the level of the entropy-based plateau. In the first step, the input histogram is divided into two separate sub-histograms, using the mean brightness as a primary threshold of total image pixels. The mentioned threshold is updated in a way that it minimizes the different values of discrete entropy between each section. Then, based on the measured plateau value, these sub-histograms are clipped to prevent unnecessary enhancement. Finally, the image intensity is stretched based on the cumulative distribution function. Laboratory results show that this method gives better outcomes of enhancement, especially in the presence of noise, compared to some two-section methods of preserving of brightness of the histogram equalization.