点云直方图在多波束条带自动匹配中的应用

Application of point cloud histogram in automatic matching of multi-beam strips

多波束测深是一种广泛使用的水下地形探测方式。当前多波束数据处理技术日臻完善,但是多波束条带间自动匹配仍存在较多问题。针对水下复杂环境、多波束自动匹配效果不佳的问题,采用点云直方图(point feature histograms,PFH)自动匹配算法,对条带点云进行自动匹配。因直方图所在的高维超空间为特征描述提供一类量化信息,对点云对应曲面的多维姿态具有鲁棒性和适用性。因此,在多波束自动匹配算法中采用PFH算法。实验数据由6205侧扫多波束测深系统获取,并对实验数据采用随机抽样一致算法(random sample consensus,RANSAC)进行定性定量分析,验证本文算法的优势,并分析相关不足。

Multi-beam sounding is a widely used method of underwater terrain detection.There are still many problems in automatic matching between multi-beam strips although the current multi-beam data processing technology is improving gradually.There is a lot of noise during the transmission of acoustic signals in view of the complex underwater environment.Point cloud histogram(Point Feature Histograms,PFH for short)automatic matching algorithm is used to automatically match strip point clouds in this paper.It is robust and applicable to the multi-dimensional pose of the corresponding surface of the point cloud because the highdimensional hyperspace where the histogram is located provides a type of quantitative information for feature description.Therefore,this article uses the PFH algorithm in the multi-beam automatic matching algorithm.The experimental data is acquired by the 6205 side-scan multi-beam sounding system,and the experimental data is analyzed qualitatively and quantitatively using the Random Sample Consensus(RANSAC)algorithm to verify the advantages of the algorithm in this paper and analyze related shortcomings.