地球信息科学学报 ›› 2018, Vol. 20 ›› Issue (4): 422-429.doi: 10.12082/dqxxkx.2018.170486

• 全国激光雷达大会特约稿件 • 上一篇    下一篇

稳健的回光反射平面靶心定位算法

付永健(), 李宗春*(), 何华   

  1. 信息工程大学地理空间信息学院,郑州 450001
  • 收稿日期:2017-10-17 修回日期:2018-01-15 出版日期:2018-04-20 发布日期:2018-04-26
  • 通讯作者: 李宗春 E-mail:15517505913@163.com;13838092876@139.com
  • 作者简介:

    作者简介:付永健(1993-),男,硕士生,主要从事精密工程测量、点云数据三维重建研究。E-mail: 15517505913@163.com

  • 基金资助:
    航天器高精度测量联合实验室基金资助项目(201501)

A Robust Algorithm of Retro-reflective Planar Target Center Positioning

FU Yongjian(), LI Zongchun*(), HE Hua   

  1. Institute of Geography Space Information, Information Engineering University, Zhengzhou 450001, China
  • Received:2017-10-17 Revised:2018-01-15 Online:2018-04-20 Published:2018-04-26
  • Contact: LI Zongchun E-mail:15517505913@163.com;13838092876@139.com
  • Supported by:
    The Foundation of High-Precision Measuring Joint Laboratory for Spacecraft, No.201501.

摘要:

针对因回光反射平面标靶点云数据缺失或冗余而难以准确计算靶心坐标的问题,本文提出一种基于距离标靶重心最远点的边缘点提取和靶心定位算法。首先,进行点云数据预处理,先人工大概选取出标靶点云所在位置,并根据回光反射强度信息提取出标靶点云,对标靶点云进行粗差剔除、投影以及坐标旋转等工作;然后,进行边缘点提取,应用所提的边缘点提取算法对投影到二维平面的标靶点云进行边缘点提取;最后,进行靶心定位,先应用抗差最小二乘对边缘点进行拟合计算圆心坐标,然后将其旋转回三维空间作为靶心坐标计算值。实验结果表明,本文提出的边缘点提取算法能高效、准确地提取出标靶边缘点,比文献[12]中的边缘点提取算法节约了大量时间,并且应用所提取出的边缘点能稳健地计算出靶心坐标,与基准值的偏差在亚毫米以内,优于文献[11]、[12]算法靶心计算精度,有效地解决了残缺或冗余的回光反射平面标靶点云靶心定位问题。

关键词: 回光反射, 平面标靶, 边缘点提取, 抗差最小二乘, 平面拟合

Abstract:

To solve the problem of calculating the center of retro-reflective planar target when the point clouds are deficient or redundant, an algorithm of extracting edge points and calculating target center is proposed. The algorithm includes three steps: (1) point clouds preprocessing; (2) edge points extracting; (3) target center calculating. In step (1), the rough region of the target points is artificially segregated from the points scanned by the laser scanner first. Then, the target points are accurately extracted from the rough region according to the intensity of return light. The noise points are removed from the target points to get the high-quality target point clouds. Finally, the high-quality target point clouds are projected into a plane, called the best fitting plane, and then the plane is rotated to be parallel with the XOY coordinate plane. In step (2), the barycenter of the target point clouds is calculated, and then all the points are translated to a new coordinate plane with the barycenter as its origin. The new coordinate plane is divided into several fan-shaped regions. The point is regarded as the edge one only when it is farthest away from the origin in one region. In step (3), the equation of the target circle is calculated by fitting the edge points, using the robust least square method. The fitting circle center is rotated back to 3D space used for target point cloud. The resulted circle center in 3D space is regarded as the estimated value of the planar target center. In order to test the effectiveness of the proposed algorithm, three tests were conducted. Firstly, the target center of high-quality target point clouds was separately calculated by the proposed algorithm and centroid method, and the accuracy of target center locations was compared. Secondly, the edge points were extracted by the proposed algorithm and the method in Ref. [12], and the time efficiency of the algorithms was compared. Thirdly, the center of low-quality target point clouds is calculated by the proposed algorithm, and methods introduced in Ref. [11] and Ref. [12], and the bias and location accuracy from these methods were compared. The experimental results show that the proposed algorithm of extracting the edge points can get good results in shorter computing time than that by Ref. [12] method. And the proposed algorithm can quickly and accurately calculate the target center, and the location accuracy is better than 1mm, better than that of Ref. [11] and Ref. [12] method. The proposed method is effective and practical.

Key words: retro-reflective, planar target, edge points extracting, robust least square, plane fitting