Underwater images often suffer from low contrast, blurring, and color distortion due to light absorption and scattering, limiting their usefulness in fields like robotics and computer vision. To address these issues, we propose a restoration method based on a revised underwater imaging degradation model, using different attenuation coefficients for direct and backscattering. We adopt linear fitting for small depth distances to improve computational efficiency and propose a statistical method for depth prior to enhance depth estimation accuracy. Our color correction preprocessing removes blue-green bias, solving color shift problems. The proposed method shows a 15.8% improvement in performance metrics compared to the conventional methods and operates 5–10 times faster.