Abstract: Aiming at the phenomena of color distortion, low contrast and image blurring in underwater images, an underwater enhancement algorithm combining guided filtering with adaptive operators is proposed. Firstly, according to the difference in the light absorption of the water body, the adaptive operator color compensation of the three channels of the underwater image is carried out, and the three channels are fused to obtain the underwater image after color compensation, which effectively improves the color authenticity of the underwater image. Then the underwater image is put into the Retinex model with improved guided filtering, which effectively removes the blurring phenomenon caused by the underwater image. Finally, the three weights of the enhanced underwater image are calculated separately, and multi-scale fusion is carried out according to the three weights to obtain the final enhanced underwater image. Different underwater images are selected for objective evaluation experiments and subjective evaluation experiments. Compared with the underwater image enhancement algorithms at this stage, the experimental results show that the proposed algorithm has good effects in correcting the color of underwater images and enhancing the contrast of underwater images, which conforms to the visual characteristics of the human eye, and the visual effect is better than the existing underwater image enhancement algorithms.

Key words: image processing, underwater image enhancement, adaptive operator, guide filtering

摘要： 针对水下图像存在的颜色失真、对比度低及图像模糊等现象，提出一种结合导向滤波与自适应算子的水下增强算法。首先根据水体对光线吸收的差异，对水下图像的三通道进行自适应算子颜色补偿，融合三通道，得到颜色补偿后的水下图像，有效提升水下图像色彩真实性；再将水下图像放入导向滤波改进的Retinex模型中，有效去除水下图像产生的模糊现象；最后分别计算增强后水下图像的三种权重，根据三种权重进行多尺度融合，获得最终增强后的水下图像。选取不同的水下图像进行客观评价实验与主观评价实验，再与现阶段的水下图像增强算法进行对比，实验结果说明该算法在修正水下图像颜色及增强水下图像对比度等方面具有较好的效果，符合人眼视觉特征，视觉效果优于现有的水下图像增强算法。

关键词: 图像处理, 水下图像增强, 自适应算子, 导向滤波