基于三维点云的结构面产状获取方法研究  

Method of Obtaining Structural Plane Occurence Based on Three-dimensional Point Cloud

作　　者：冯文凯[1] 曾唯恐 程柯力 易小宇[1] 焦隆新 FENG Wen-kai;ZENG Wei-kong;CHENG Ke-li;YI Xiao-yu;JIAO Long-xin(State Key Laboratory of Geohazard Prevention and Geoenvironment Protection,Chengdu University of Technology,Chengdu 610059,China)

机构地区：[1]成都理工大学地质灾害防治与地质环境保护国家重点实验室,成都610059

出　　处：《长江科学院院报》2021年第9期128-132,140,共6页Journal of Changjiang River Scientific Research Institute

基　　金：国家自然科学基金项目(41977252);2018年度交通运输行业重点科技项目(2018-ZD5-029)。

摘　　要：为快速从地质结构面三维点云数据中提取产状信息,基于Python程序设计语言,编程实现了一套自动拟合平面并计算结构面产状的算法。首先,介绍了最小二乘法和主成分分析法2种算法原理和求解平面方程思路;其次,利用Python语言分别设计实现了上述算法,并引入奇异值分解帮助求解主成分向量,给出了关键代码和程序流程;最后,对2种算法进行对比和误差分析,并将平面方程转换为产状信息。将该方法应用于国际公开试验数据,人工截取指定结构面产状,计算结果平均值与实际值相比<1°,最大不超过2°;无监督聚类分割生成的不规则结构面产状计算结果平均值与实际值相比<4°,最大不超过8°,且主成分分析法误差更小。结果表明,该方法精确度高,使用简便,满足工程实际需要。To extract rapidly the occurrence information from three-dimensional point cloud data of geological structural plane,we completed a set of algorithms that automatically fit the plane and calculate the structural plane occurrence by programming using Python.First of all,we expounded the principles of least squares and principal component analysis as well as the solution of plane equations;secondly,we designed the above two algorithms using Python language,and introduced singular value decomposition to help solve the principal component vector,and gave the key code and program flow;finally,we compared the two algorithms and analyzed their errors,and converted the plane equation into occurrence information.We then applied the present method to international public experimental data.Results manifested that for specified structural planes manually intercepted,the calculation error of structural plane occurrence was less than 1°on average compared with the actual value,not exceeding 2°;for irregular structural planes generated by unsupervised clustering segmentation,the calculation error was less than 4°on average compared with the actual value,not exceeding 8°.The error of principal component analysis method was even smaller.The results demonstrated that the present method is of high accuracy and convenience,and hence meeting practical engineering requirements.

关 键 词：结构面产状 三维点云数据 平面拟合 最小二乘法 主成分分析法 Python程序 

分 类 号：P642[天文地球—工程地质学]