摘要
针对点云平面拟合中存在粗差及异常值等问题,对结合特征值法的随机抽样一致性(random sample consensus,RANSAC)平面拟合算法进行了改进。该方法以RANSAC算法为基础,结合特征值法,利用点到平面模型距离的标准偏差来自动选取阈值t,通过阈值t检测并剔除异常数据点,达到获得理想平面拟合参数的目的。用改进的算法和传统的特征值法分别对点云数据进行处理,结果表明,改进的算法适用于存在误差和异常值的点云数据拟合,能稳定地获得较好的平面参数估值,具有较强的鲁棒性。
In the process of plane fitting of point clouds,there are some grosserrors and outliers.In order to overcome this shortcoming,a robust plane fitti-ng method based on RANSAC(RANdom SAmple Consensus)was improved in the paper.It combined with eigenvalue method and obtained some ideal plane fitting paramete-rs through setting threshold value t to find and remove the gross errors and o-utliers.Threshold value is automatically selected by the standard deviation of the distance from the point to the plane model.The improved algorithm was used to do plane fitting with the measured data,and result showed that compared witheigenvalue method,the improved method could do well with the point clouds containing errors and outliers,thus be a robust plane fitting algorithm.
出处
《北京测绘》
2016年第2期73-75,79,共4页
Beijing Surveying and Mapping
