Underwater Image Enhancement by Illumination Map Estimation and Adaptive High-Frequency Gain
Zheng Cui, Chunxi Yang, Sen Wang, Xian Wang, Hao Duan, Jing Na · IEEE Sensors Journal · 2024
Images captured in underwater environments are typically affected by nonuniform illumination, which degrades their quality and consequently restricts their practical application in visual tasks. To improve the quality of underwater images, we propose an underwater image enhancement method based on the brightness correction and an adaptive gain control function. Specifically, we first utilize color transfer image and the maximum attenuation map to derive the color-corrected image. Then, we tackle the challenge of nonuniform illumination in underwater images by imposing a structure weight matrix. The integral map is also utilized to compute the low-frequency information of the image. Finally, an adaptive gain control function is proposed to enhance the high-frequency information of the image adaptively. Comprehensive experiments with both qualitative and quantitative analysis are implemented on the OceanDark underwater dataset, and the results demonstrate that proposed method outperforms some mainstream methods. Further experiments on a real-world industrial underwater environment also validate that the proposed method can effectively improve the accuracy of binocular camera calibration and distance measurement.