Please wait a minute...
Chinese Journal of Engineering Design  2021, Vol. 28 Issue (2): 155-162    DOI: 10.3785/j.issn.1006-754X.2021.00.029
Design for Quality     
Automatic repair of scattered point cloud hole based on GA-BP neural network
WANG Chun-xiang, HAO Lin-wen, WANG Yao, ZHOU Guo-yong, JI Kang-hui, LIU Liu
School of Mechanical Engineering, Inner Mongolia University of Science and Technology, Baotou 014010, China
Download: HTML     PDF(3189KB)
Export: BibTeX | EndNote (RIS)      

Abstract  Point cloud hole repair is a key technology in the point cloud data processing, which directly affects the quality and integrity of the point cloud. The BP (back propagation) neural network optimized by genetic algorithm (GA)(GA-BP neural network) is a good repair method for scattered point cloud holes. However, multiple steps in the traditional repair method of scattered point cloud holes based on GA-BP neural network need to be completed through the human-computer interaction with the help of reverse software, which leads to a tedious and time-consuming process. Therefore, an automatic repair method of scattered point cloud holes based on GA-BP neural network was proposed. Through the computer programming, the hole identification, hole region interpolation and hole repair were combined to realize the automatic repair from the incomplete point cloud model to complete point cloud model without complicated human-computer interaction and data conversion. The experimental results show that the proposed method can effectively avoid the data distortion caused by the data conversion, reduce the workload of human-computer interaction and repair point cloud holes conveniently and efficiently. The density of the repaired point cloud is uniform, which is of great significance for improving the repair efficiency and quality of point cloud holes.

Received: 11 June 2020      Published: 28 April 2021
CLC:  TP 391.4  
Cite this article:

WANG Chun-xiang, HAO Lin-wen, WANG Yao, ZHOU Guo-yong, JI Kang-hui, LIU Liu. Automatic repair of scattered point cloud hole based on GA-BP neural network. Chinese Journal of Engineering Design, 2021, 28(2): 155-162.

URL:

https://www.zjujournals.com/gcsjxb/10.3785/j.issn.1006-754X.2021.00.029     OR     https://www.zjujournals.com/gcsjxb/Y2021/V28/I2/155


基于GA-BP神经网络的散乱点云孔洞自动修补

点云孔洞修补作为点云数据处理中的关键技术,直接影响点云的质量和完整性。利用遗传算法(genetic algorithm,GA)优化的BP(back propagation,反向传播)神经网络(简称GA-BP神经网络)是一种修补效果较好的散乱点云孔洞修补方法。但基于GA-BP神经网络的散乱点云孔洞传统修补方法的多个步骤需借助逆向软件通过人机交互的方式完成,导致修补过程繁琐且耗时较长。为此,提出了一种基于GA-BP神经网络的散乱点云孔洞自动修补方法。通过计算机编程将孔洞识别、孔洞区域插值和孔洞修补相结合,实现从残缺点云模型直接到完整点云模型的自动修补,无须进行复杂的人机交互和数据转换。实验结果表明,所提出的方法可有效避免因数据转换而造成的数据失真,减少了人机交互工作量,方便而高效地修补了散乱点云的孔洞,且得到的修补点云密度均匀,这对提高点云孔洞修补效率和质量具有重要意义。
[1] LAI Quan-bao, TAO Qing, HU Yu-ge, MENG Qing-feng. Multi-gesture accurate recognition based on artificial fish swarm algorithm-extreme learning machine[J]. Chinese Journal of Engineering Design, 2021, 28(6): 671-678.
[2] WANG Wei, ZHANG Kai-yan, LIU Ya-chuan, HUANG Yu-chun. Research on two-step alignment method for 3-D point cloud and its application[J]. Chinese Journal of Engineering Design, 2020, 27(5): 560-567.
[3] SONG Guang-yue, SONG Zhi-bin, XIANG Zhong-xia. Gait phase recognition under proportion-uncontrolled body weight support based on plantar pressure sensor[J]. Chinese Journal of Engineering Design, 2019, 26(3): 260-266.