作者：张越¹, 翟福琪², 蔡孙宝³, 陈鹏^3,4

作者单位：1. 成都飞机工业（集团）有限责任公司，四川 成都 610092;
2. 华中科技大学机械科学与工程学院，湖北 武汉 430074;
3. 西南交通大学机械工程学院，四川 成都 611756;
4. 四川省机械研究设计院（集团）有限公司，四川 成都 610063

关键词：结构光;点云;特征提取;曲面重建

摘要：

如何准确而高效地获取、处理和重建物体的三维点云数据是制约三维检测技术发展的关键性问题。该文主要以植物叶片为载体，对三维点云的数据获取、特征提取以及曲面重建技术进行研究。针对三维点云的获取问题，采用面结构光三维扫描仪采集植物叶片表面的三维点云数据，进而开展点云数据的预处理，包括去噪、精简和平滑。采用基于曲率和基于点云距离的特征提取方法分别对叶片叶脉信息进行提取，并对两种方法进行对比；引入基于边界点邻域特征的边界提取算法得到植物叶片的边缘轮廓信息。讨论三维点云曲面重建算法及其原理，分析贪婪投影三角化算法和泊松曲面重建算法的效果。结果表明，该文所提出的特征提取和三维重建方法，在合适的参数设置下能够在两种不同的植物叶片上得到一致的效果。

Feature extraction and 3D reconstruction of plant leaf based on point cloud data
ZHANG Yue¹, ZHAI Fuqi², CAI Sunbao³, CHEN Peng^3,4
1. Chengdu Aircraft Industrial (Group) Co., Ltd., Chengdu 610092, China;
2. School of Mechanical Science and Engineering, Huazhong University of Science and Technology, Wuhan 430074, China;
3. College of Mechanical Engineering, Southwest Jiaotong University, Chengdu 611756, China;
4. Sichuan Provincial Machinery Research & Design Institute (Group) Co., Ltd., Chengdu 610063, China
Abstract: How to accurately and efficiently acquire, process and reconstruct 3D point cloud data of objects is a key issue that restricts the progress of 3D inspection technology. In this paper, the plant leaf is taken to study the acquisition, feature extraction and surface reconstruction of 3D point cloud. For the acquisition of 3D point cloud, the surface structured light 3D scanner is used to collect 3D point cloud data on the surface of plant leaves, and then the point cloud data is preprocessed, including denoising, reducing and smoothing. The feature extraction method based on curvature and point cloud distance is used to extract the blade vein information, and the two methods are compared. The edge contour information of plant leaves is extracted by the boundary extraction algorithm based on the features of the boundary point. The 3D point cloud surface reconstruction algorithm and its principle are studied. The greedy projection triangulation algorithm and the Poisson surface reconstruction algorithm are used to analyze the reconstruction effect. The results show that the proposed feature extraction and surface reconstruction methods could obtain consistent effects on the two different plant leaves if they have proper parameter settings.
Keywords: structured light;point cloud;feature extraction;surface reconstruction
2021, 47(8):6-12  收稿日期: 2020-12-01;收到修改稿日期: 2021-01-13
基金项目: 四川省省级工业发展专项资金项目（2020NJ002）；四川省科技计划重点研发项目（2020YFS0086）
作者简介:
参考文献
[1] 陆红红, 殷鸣, 谢罗峰, 等. 基于面结构光的叶片三维重构技术研究[J]. 中国测试, 2019, 45(2): 134-138
[2] LANGE C, POLTHIER K. Anisotropic smoothing of point sets[J]. Computer Aided Geometric Design, 2005, 22(7): 680-692
[3] BERKMANN J, CAELLI T. Computation of surface geometry and segmentation using covariance techniques[C]//IEEE Trans Pattern Analysis and Machine Intelligence, 1994: 1114-1116.
[4] 庞旭芳, 庞明勇, 肖春霞. 点云模型谷脊特征的提取与增强算法[J]. 自动化学报, 2010, 36(8): 1073-1078
[5] HOPPE H, DEROSE T, DUCHAMP T. Surface reconstruction from unorganized points[J]. Computer Graphics(ACM), 1992, 26(2): 71-78
[6] KAZHDAN M, BOLITHO M, HOPPE H. Poisson surface reconstruction[C]//Eurographics Symposium on Geometry Processing, 2006: 61-70.
[7] WANG L, LU L, JIANG N. A study of leaf modeling technology based on morphological features[J]. Mathematical and Computer Modeling, 2011, 54(3): 1107-1114
[8] MILENKOVI M, EYSN L, HOLLAUS M, et al. Modeling the tree branch structure at very high resolution[C]//Silvi Laser, 2012: 1-8.
[9] 孙智慧, 陆声链, 郭新宇, 等. 基于点云数据的植物叶片曲面重构方法[J]. 农业工程学报, 2012, 28(3): 184-190
[10] 万军, 鞠鲁粤. 逆向工程中数据点云精简方法研究[J]. 上海大学学报, 2004, 2(1): 26-29
[11] 代星, 崔汉国, 胡怀宇. 基于曲率特征的点云快速简化算法[J]. 计算机应用, 2009, 29(11): 3030-3032
[12] 柯映林, 范树迁. 基于点云的边界特征直接提取技术[J]. 机械工程学报, 2004, 40(9): 116-120
[13] 周春令. 基于结构光的三维表面重建技术研究[D]. 杭州: 浙江大学, 2018.
[14] DAVID C, FRANK D. A greedy Delaunay-based surface reconstruction algorithm[J]. The Visual Computer, 2004, 20(1): 4-16
[15] KAZHDAN M, BOLITHO M, HOPPE H. Screened poisson surface reconstruction[J]. ACM Transactions on Graphics, 2013, 32(3): 61-70