表面波和Retinex结合的水声图像增强方法

Image Enhancement for Sonar Images Based on Surfacelet and Multi-scale Retinex

水下复杂信道环境以及各种干扰的存在使得声呐图像分辨率偏低,边缘细节不便于识别处理,传统增强算法大多直接进行增强处理而较少考虑去噪过程.针对这一问题,该文分析了表面波分解和视网膜皮层图像增强原理,阐述了两者结合的可能性,并在此基础上提出了一种先采用表面波结合自适应阈值去噪处理,再进行多尺度Retinex增强的图像处理方法.在仿真实验中将该方法与其他图像去噪增强算法分别进行比较,结果表明该方法在边缘细节保持及颜色保真方面具有优势,能够在获得更好的视觉结果的同时控制算法复杂度,有利于后续图像处理.

Due to complicated underwater channels and various interferences, sonar image resolution is low,leading to difficulties in identifying edge details. Many enhancement algorithms can improve image resolution without de-noising. To this end, the paper proposes an improved enhancement algorithm, which first performs de-noising using surfacelet transform with adaptive threshold, and then carries out multi-scale retinex enhance- ment. Simulation results show that the proposed algorithm has advantages in preservation of edge details color information. It helps subsequent image processing.