Histogram equalization using a selective filter

Abstract Many popular modern image processing software packages implement a naïve form of histogram equalization. This implementation is known to produce histograms that are not truly uniform. While exact equalization techniques exist, these may undesirable artifacts in some scenarios. In this paper we consider the link between established continuous theory for global and its discrete implementation, formulate novel technique builds upon considerably improves approach. We show can linearly interpolate cumulative distribution low-bit by approximately dequantizing intensities using selective box filter. helps distribute more evenly. The proposed algorithm subsequently evaluated compared with existing works literature. find method capable producing an equalized has high entropy, while distances similar preserved. described approach implications on several related problems, e.g., edge detection.