Adaptive histogram equalization framework based on new visual prior and optimization model

Abstract

Histogram Equalization (HE) algorithm remains one of the research hotspots in the field of image enhancement due to its computational simplicity. Despite numerous improvements made to HE algorithms, few can comprehensively account for all major drawbacks of HE. To address this issue, this paper proposes a novel histogram equalization framework, which is an adaptive and systematic resolution. Firstly, a novel optimization mathematical model is proposed to seek the optimal controlling parameters for modifying the histogram. Additionally, a new visual prior knowledge, termed Narrow Dynamic Prior (NDP), is summarized, which describes and reveals the subjective perceptual characteristics of the Human Visual System (HVS) for some special types of images. Then, this new knowledge is organically integrated with the new model to expand the application scope of HE. Lastly, unlike common brightness preservation algorithms, a novel method for brightness estimation and precise control is proposed. Experimental results demonstrate that the proposed equalization framework significantly mitigates the major drawbacks of HE, achieving notable advancements in striking a balance between contrast, brightness and detail of the output image. Both objective evaluation metrics and subjective visual perception indicate that the proposed algorithm outperforms other excellent competition algorithms selected in this paper.