基于改进γ-CLAHE算法的水下机器人图像识别

水体及悬浮粒子对光的吸收、折射及反射导致水下图像对比度低及细节模糊，单一图像增强算法难以适用于水下复杂环境识别.为了解决该问题，提出基于小波变换和改进的γ-CLAHE相融合的图像增强算法.通过快速中值滤波去除图像中噪声，向CLAHE算法中加入自适应伽马变换，解决CLAHE算法处理水下图像色彩失真，丢失孤立点、细线，画面突变等问题. 利用改进的γ-CLAHE算法处理小波变换分解后的低频部分，增强图像并加快运行速度. 通过小波逆变换将γ-CLAHE算法处理后的低频部分和双边滤波处理后的高频部分相融合，得到最终的增强图像. 将实验图像同传统CLAHE、Retinex、Singh融合算法的处理图像进行对比，验证本研究算法在水下图像处理方面的有效性和优越性.

Recognition of images for underwater vehicle based on improved γ-CLAHE algorithm

The absorption, refraction and reflection of light by water and suspended particles lead to low contrast and blurred details of underwater images. Therefore, it is difficult to apply a single image enhancement algorithm to the recognition of complex underwater environments. An enhancement algorithm based on wavelet transform and an improved γ-CLAHE algorithm was proposed to solve this problem. Firstly, fast median filter was used to remove the noise in the image, and adaptive gamma transform was added to CLAHE to solve the problems of color distortion and loss of details information such as isolated points, thin lines and sudden changes in the underwater image. Secondly, the improved γ-CLAHE method was used to process the low frequency part after wavelet transform decomposition to enhance the image and speed up the algorithm. Then, the wavelet inverse transform was used to get the final enhanced image by fusing the low-frequency part processed by the γ-CLAHE algorithm and the high-frequency part processed by bilateral filtering. Finally, the final image was compared with the images processed by traditional CLAHE, Retinex, and Singh’s fusion algorithm, verifying the effectiveness and superiority of the proposed algorithm in the underwater image processing.