Underwater images are often compromised by light absorption and scattering, leading to diminished contrast, color distortion, and blurred details, which significantly impede the effectiveness of underwater target detection and localization algorithms. To address the issue, an underwater image enhancement method based on a physical model is proposed in this paper. First, the method integrates gradient computation into conventional depth estimation algorithms to obtain more accurate depth values. Second, a quadtree decomposition algorithm is employed to refine the estimation of background light. Subsequently, histogram equalization is utilized to enhance image contrast, thereby improving overall image quality. The experimental results show that the proposed method outperforms several other classical image enhancement methods. Meanwhile, this method has been applied in pool experiments, which has improved the success rate of target detection.