顾及海底纹理信息熵的侧扫图像修复方法

Texture restoration method of side scan sonar image based on gray entropy

针对侧扫声纳图像由于成像机理、测量及数据处理方法造成的图像异常和缺失问题,提出了一种基于图像灰度熵的纹理修复方法。首先开展了异常区探测方法研究,提出了联合拖鱼高度和图像灰度梯度的纹理异常区探测方法和基于图像分块比较的缺失区检测方法;然后提出了一种联合图像灰度熵最优权计算和样本相似块搜寻的图像异常区修复方法。实验表明,该方法可以有效地恢复缺失和失真区的纹理,保证了修复区海底地貌纹理结构的合理性,改善了侧扫声纳图像质量。

In this paper,we conduct a research on the repair of abnormal regions of side-scan sonar images due to weak echo intensity values caused by radiation factors,the absence of stitching regions caused by inaccurate seabed line tracking and the absence of textures caused by target shadows,and propose a gray-scale entropy-based texture repair method.The method combines the algorithms of automatic search of the restored region,calculation of the optimal weight based on gray-scale entropy and search of similar regions of the image.The experiments show that the method can effectively recover the texture of the missing and distorted regions,ensure the similarity of the texture structure of the restored regions,and greatly improve the image quality.