Hole Identification of Complex Parts for Automatic Repair
-
摘要: 依据样点的邻域在其局部微切平面投影的分布情况,提出了一种针对采样密度不均、几何形状复杂、孔洞面积大小不一的散乱点云模型的孔洞识别方法。通过使用NP邻域,使得位于密度过渡区域的点被准确分类;由于样点的邻域可能跨过多个面,常规PCA方法估算的点云法矢不准确,从而导致模型尖锐位置上的点误判,通过引入距离权重,保证了局部微切平面计算的准确性;针对邻域点数k取值较少,检测结果中存在较多噪声点,而k值较大又会覆盖模型中较小的孔洞,通过邻域支持的方法,有效地检测出模型中的小面积孔洞;为有利于自动化修补孔洞,文中采用划分空间栅格聚类的方法确定孔洞位置及数量,避免了点与点之间距离的反复计算,加快了聚类速度。实验结果表明,该方法能有效检测模型中面积大小不一的孔洞,得到的检测结果噪声点少,孔洞轮廓清晰。Abstract: According to the distribution of the neighborhood of the sample point projected in its local micro-cut plane, a hole identification method for the scattered point cloud model with uneven sampling density, complex geometry and different hole area is proposed. With the NP neighborhood, the points located in the density transition area are accurately classified; since the neighborhood of the sample points may span multiple faces, the point cloud normal vector estimated by the PCA (Principle analysis) method is inaccurate, resulting in a point error in the sharp position of the model. Judging, by introducing the distance weight, the accuracy of the local micro-cut plane calculation is guaranteed; for the neighborhood point number k is less, there are more noise points in the detection result, and the larger k value will cover the smaller of the model. Holes, through the method of neighborhood support, effectively detect small-area holes in the model. In order to facilitate the automatic repair of holes, the method of dividing spatial grid clustering is used to determine the position and number of holes, which avoids the repeated calculation of the distance between points and accelerates the clustering speed. The experimental results show that the method can effectively detect the holes with different sizes in the model, and the obtained detection results have fewer noise points and clear hole contours.
-
Key words:
- point cloud /
- fitting plane /
- space grid /
- edge detection
-
表 1 检测结果与运行时间对比(
$k = 20$ ,$s = 3$ )算法 检测结果 误判点数 误判比率/% 运行时间/s $N_P^k$邻域 11 914 11 532 96.79 1.74 ${N_P}$邻域 2 541 2 153 84.73 17.61 本文方法 444 61 13.74 18.03 -
[1] 曾露露, 盖绍彦, 达飞鹏, 等. 基于从运动中恢复结构的三维点云孔洞修补算法研究[J]. 光学学报, 2018, 38(6): 0610002 doi: 10.3788/AOS201838.0610002ZENG L L, GAI S Y, DA F P, et al. Hole filling algorithm of three-dimensional point cloud based on structure from motion[J]. Acta Optica Sinica, 2018, 38(6): 0610002 (in Chinese) doi: 10.3788/AOS201838.0610002 [2] 程效军, 方芳. 基于形态学的散乱点云轮廓特征线提取[J]. 同济大学学报, 2014, 42(11): 1738-1743 doi: 10.11908/j.issn.0253-374x.2014.11.018CHENG X J, FANG F. Morphology-based scattered point cloud contour extraction[J]. Journal of Tongji University, 2014, 42(11): 1738-1743 (in Chinese) doi: 10.11908/j.issn.0253-374x.2014.11.018 [3] ZOLANVARI S M I, LAEFER D F, NATANZI A S. Three-dimensional building façade segmentation and opening area detection from point clouds[J]. ISPRS journal of Photogrammetry and Remote Sensing, 2018, 143: 134-149 doi: 10.1016/j.isprsjprs.2018.04.004 [4] JUN Y. A piecewise hole filling algorithm in reverse engineering[J]. Computer-Aided Design, 2005, 37(2): 263-270 doi: 10.1016/j.cad.2004.06.012 [5] 孙殿柱, 刘华东, 史阳, 等. 基于核密度估计的散乱点云边界特征提取[J]. 农业机械学报, 2013, 44(12): 275-279, 268 doi: 10.6041/j.issn.1000-1298.2013.12.046SUN D Z, LIU H D, SHI Y, et al. Boundary feature abstraction of unorganized points based on kernel density estimation[J]. Transactions of the Chinese Society for Agricultural Machinery, 2013, 44(12): 275-279, 268 (in Chinese) doi: 10.6041/j.issn.1000-1298.2013.12.046 [6] 陈义仁, 王一宾, 彭张节, 等. 一种改进的散乱点云边界特征点提取算法[J]. 计算机工程与应用, 2012, 48(23): 177-180, 190 doi: 10.3778/j.issn.1002-8331.2012.23.040CHEN Y R, WANG Y B, PENG Z J, et al. Improved algorithm for extraction of boundary characteristic point from scattered point cloud[J]. Computer Engineering and Applications, 2012, 48(23): 177-180, 190 (in Chinese) doi: 10.3778/j.issn.1002-8331.2012.23.040 [7] 刘庆, 章光, 陈西江. 融合改进场力和判定准则的点云特征规则化[J]. 中国激光, 2019, 46(4): 0404009 doi: 10.3788/CJL201946.0404009LIU Q, ZHANG G, CHEN X J. Point cloud feature regularization based on fusion of improved field force and judging criterion[J]. Chinese Journal of Lasers, 2019, 46(4): 0404009 (in Chinese) doi: 10.3788/CJL201946.0404009 [8] NI H, LIN X G, NING X G, et al. Edge detection and feature line tracing in 3D-point clouds by analyzing geometric properties of neighborhoods[J]. Remote Sensing, 2016, 8(9): 710 doi: 10.3390/rs8090710 [9] BENDELS G H, SCHNABEL R, KLEIN R. Detecting holes in point set surfaces[J]. Journal of WSCG, 2006, 14: E43 [10] GUMHOLD S, WANG X L, MACLEOD R S. Feature extraction from point clouds[C]//Proceedings of the 10th International Meshing Roundtable. California: Sandia National Laboratories, 2001: 293-305 [11] MEDEROS B, VELHO L, DE FIGUEIREDO L H. Robust smoothing of noisy point clouds[C]// Proceedings of the SIAM Conference on Geometric Design and Computing. Seattle: SIAM, 2003: 405-416 [12] HOPPE H, DEROSE T, DUCHAMP T, et al. Surface reconstruction from unorganized points[J]. ACM SIGGRAPH Computer Graphics, 1992, 26(2): 71-78 doi: 10.1145/142920.134011 [13] 王小超. 三维数字几何处理中特征提取与孔洞修补研究[D].大连: 大连理工大学, 2014WANG X C. Studies on feature extraction and hole-filling in 3D digital geometry processing[D]. Dalian: Dalian University of Technology, 2014 (in Chinese) [14] 孙殿柱, 范志先, 李延瑞. 散乱数据点云边界特征自动提取算法[J]. 华中科技大学学报, 2008, 36(8): 82-84SUN D Z, FAN Z X, LI Y R. Automatic extraction of boundary characteristic from scatter data[J]. Journal of Huazhong University of Science and Technology, 2008, 36(8): 82-84 (in Chinese) [15] 薛连杰, 齐臣坤, 张彪, 等. 基于3维点云欧氏聚类和RANSAC边界拟合的目标物体尺寸和方位识别[J]. 机械设计与研究, 2018, 34(5): 44-48, 53XUE L J, QI C K, ZHANG B, et al. Object size and orientation recognition based on 3D point cloud euclideam clustering and RANSAC boundary fitting[J]. Machine Design & Research, 2018, 34(5): 44-48, 53 (in Chinese)