Aiming at the two degradation problems of Imaging color distortion caused by light attenuation and uneven illumination caused by artificial light source in underwater low-illumination environment,an underwater low-illumination image enhancement network based on color correction and brightness estimation named LUcobe is proposed.LUcobe is composed of color correction sub network and brightness estimation sub network.The color correction sub network extracts the most discriminative color features through the dual color space and combines with the attention mechanism,and integrates and highlights them adaptively.The brightness estimation sub network uses the encoder-decoder structure and dilate convolution to globally estimate the illumination of the image.Finally,combines the output of the brightness estimation sub network with the output of the color correction sub network through selective channel module to achieve image enhancement.The experimental results show that LUcobe can effectively improve the uneven illumination and color deviation of the underwater low illumination image,and achieve better effect in visual effect and objective evaluation indexes (PSNR,SSIM,MS-SSIM,MSE) compared to other typical underwater image enhancement methods.
万方数据