Underwater image enhancement via histogram similarity-oriented color compensation complemented by multiple attribute adjustment

Abstract

AbstractUnderwater images are often influenced by color casts, low contrast, and blurred details. We observe that images taken in natural settings typically have similar histograms across color channels, while underwater images do not. To improve the natural appearance of an underwater image, it is critical to improve the histogram similarity across its color channels. To address this problem, we develop a histogram similarity-oriented color compensation method that corrects color casts by improving the histogram similarity across color channels in the underwater image. In addition, we apply the multiple attribute adjustment method, including max-min intensity stretching, luminance map-guided weighting, and high-frequency edge mask fusion, to enhance contrast, saturation, and sharpness, effectively addressing problems of low contrast and blurred details and eventually enhancing the overall appearance of underwater images. Particularly, the method proposed in this work is not based on deep learning, but it effectively enhances a single underwater image. Comprehensive empirical assessments demonstrated that this method exceeds state-of-the-art underwater image enhancement techniques. To facilitate public assessment, we made our reproducible code available at https://gitee.com/wanghaoupc/UIE_HS2CM2A.