Underwater image enhancement based on adaptive information transfer and weighted stationary wavelet perception fusion

Underwater images commonly suffer from issues like color distortion, low contrast, and blurred details. To tackle these degradation problems, we propose an underwater image enhancement (UIE) method named ATWF, which is based on adaptive information transfer and weighted stationary wavelet perception fusion. This method firstly introduces an adaptive information transfer strategy, grounded in histogram similarity cue, to effectively compensate for the attenuation channels. Subsequently, it integrates linear stretching to increase the dynamic range of color compensated image. Secondly, an adaptive gamma correction algorithm is used to enhance the overall contrast of the color-corrected image, and a multi-scale side window box filter (SWBF) technique is employed to improve its local details simultaneously. Finally, the stationary wavelet transform is used to decompose the enhanced images into low-frequency (LF) and high-frequency (HF) components. Perception fusion rules tailored to the different characteristics of each component are devised to improve the visual quality of the resultant image. Experiments on multiple public datasets show that: (1) ATWF can effectively correct color distortion; (2) it significantly improves image contrast, suppresses noise, and highlights details; (3) in terms of qualitative and quantitative evaluation, corner detection, image matching, and image segmentation, it outperforms or is comparable to other mainstream UIE methods. Additionally, ATWF demonstrates strong generalization ability in various complex degradation scenarios.