运用三维声纳图像的水底全景拼接与绘制  

Submarine global mosaicing and plotting based on 3D sonar imaging

作　　者：李志华[1,2] 李秋峦 

机构地区：[1]杭州师范大学信息科学与工程学院,杭州311100 [2]浙江大学数字技术及仪器研究所,杭州310027

出　　处：《测绘科学》2015年第6期146-150,共5页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(61001170)

摘　　要：针对海量声纳图像多帧配准引起的误差累积问题,该文提出了一种基于三维声纳图像的水底全景拼接与绘制方法。首先通过相邻两帧图像最近邻点迭代,并结合GPS、姿态仪信息和X84控制点剔除方法,减少非重叠区域控制点对的存在以提高配准精度,实现相邻帧图像精确配准。然后采用四元组参数对所有精确配准矩阵进行拟牛顿最优化处理,获取所有声纳帧之间最佳配准关系,减少多帧配准引起的累积误差,实现三维图像高精度全局配准。湖试与海试实验结果表明,该方法有效地减少了海量声纳图像多帧配准引起的累积误差,实现了三维声纳图像水底全景高精度拼接与绘制,达到了预期目标。Aiming at the cumulative error problem caused by multi-frame sonar image registration, a submarine global mosaicing and plotting method based on 3D sonar imaging was proposed. Firstly, 3D sonar imaging accurate registration was achieved by the iterated closest points and x84 control points selec-tion. And then 3D sonar imaging global registration was fulfilled by the quasi-newton optimization of accurate registration matrixes to avoid the cumulation of registration error. Experimental results in the lake and ocean showed that the system efficiently reduced the cumulative registration error of multi-frame sonar im- ages, and achieved 3D sonar data globally accurate mosaicing as well as plotting.

关 键 词：三维声纳 图像配准 图像拼接 图像绘制 

分 类 号：P229.1[天文地球—大地测量学与测量工程]