In order to improve quality of underwater images, an underwater gray image enhancement method is proposed based on controllable nuclear bilateral filtering Retinex algorithm. An improved Sigmoid function is introduced to enhance the brightness of transformed illumination images, and a bilateral filtering function with controllable kernel function is proposed to estimate the image light intensity, which overcomes edge blurring and halo artifacts problem. Illumination estimation image is corrected by using gain function. Results show that the proposed algorithm accords with human visual characteristics, which can effectively improve the contrast and resolution of underwater images.