Underwater Image Enhancement Based on Color Correction and Detail Enhancement

To solve the problems of underwater image color deviation, low contrast, and blurred details, an algorithm based on correction detail enhancement is proposed. First, improved nonlocal means denoising used to denoise image. The combination Gaussian weighted spatial distance Euclidean as index measure similarity structural blocks. can retain more edge features texture information while maintaining noise reduction ability. Then, U-Net for correction. Introducing residual structure attention mechanism into effectively enhance feature extraction ability prevent network degradation. Finally, a sharpening maximum posteriori proposed after correction, which increase detailed without expanding noise. experimental results show that has remarkable effect enhancement.