基于方向预测规则化的建筑物轮廓线提取算法  

Building Contour Extraction AlgorithmBased on Direction Prediction Regularization

作　　者：王安琪 WAN Anqi(School of Management,Dalian University of Finance and Economics,Dalian 116600)

机构地区：[1]大连财经学院管理学院,辽宁大连116600

出　　处：《常州工学院学报》2023年第5期35-40,共6页Journal of Changzhou Institute of Technology

摘　　要：设计基于方向预测规则化的建筑物轮廓线提取算法,提升轮廓线提取完整性。利用机载激光雷达采集建筑物点云数据,通过改进的散点轮廓算法在点云数据内提取建筑物轮廓点,利用道格拉斯普克算法,在轮廓点内提取关键轮廓点,采用随机抽样一致算法拟合关键轮廓点获取简化轮廓线,通过方向预测规则化算法规则化处理简化轮廓线获取最终建筑物轮廓线提取结果。实验证明:该算法可有效采集建筑物点云数据,提取建筑物轮廓点与轮廓线;在提取不同建筑物屋顶形状时,该算法轮廓线提取的完整率、点云贡献率均较高,最大偏差均较低,即轮廓线提取完整性较优、精度较高。A building contour line extraction algorithm based on direction prediction regularization is designed to enhance the integrity of contour line extraction.The airborne lidar is used to collect the building point cloud data,the building contour points are extracted within the point cloud data by the improved scatter contour algorithm,the key contour points are extracted within the contour points by using the Douglas-Peucker algorithm,the key contour points are fitted with the key contour points by using the random sample consistent algorithm to obtain the simplified contour line,and the simplified contour lines are processed by the direction prediction regularization algorithm regularly to obtain the final building contour line extraction results.Experiments have shown that the algorithm can effectively collect building point cloud data and extract building contour points and contour lines;when extracting different building roof shapes,the algorithm’s contour line extraction has a higher completeness rate and a higher point cloud contribution rate,and the maximum deviation is lower,indicating better completeness and higher accuracy in contour line extraction.

关 键 词：方向预测 规则化 建筑物 轮廓线提取 道格拉斯普克算法 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]