摘要
为了解决冷黏运动鞋自动化生产过程中鞋帮打磨轨迹难以提取的问题,提出一种利用非刚性点云配准算法提取轨迹的方法。搭建三维视觉平台分别获取鞋底侧墙与鞋帮点云,基于高斯混合模型,结合运动相干性和先验分布构建了鞋底与鞋帮点云配准概率的联合概率密度分布,并采用变分贝叶斯推断计算出最佳配准参数,进而实现两种点云间的精确配准。对配准后的点云提取边缘轨迹并聚类分割,提取的边缘轨迹即为鞋帮打磨轨迹。研究证明,该方法可有效提取运动鞋的鞋帮打磨轨迹曲线,相比依靠离线采样点提取打磨轨迹的传统方法,可大幅缩减轨迹的获取时间且精度良好。
In order to extract the upper grinding track of cold-stick athletic shoes,a non-rigid point cloud registration algorithm was proposed to extract the track.A three-dimensional vision platform was established to obtain the point clouds of the side-wall of the sole and the upper respectively.Based on the Gaussian mixture model,the joint probability density distributions of alignment probability between sole and upper point cloud were constructed by combining the motion coherence and prior distribution.The optimal registration parameters were calculated by variable Bayesian inference,so as to realize the accurate registration between two point clouds.The edge trajectory was extracted from point clouds after registration and segmented by clustering.The extracted edge trajectory was the polishing trajectory of the upper.The research shows that this method can effectively extract the trajectory curve of the upper.Compared to the traditional method of extracting grinding track based on offline sampling points,it can greatly reduce the acquisition time of track and has good accuracy.
作者
柯宇
陈玉洁
张豪
邢礼源
KE Yu;CHEN Yujie;ZHANG Hao;XING Liyuan(College of Mechanical Engineering,Donghua University,Shanghai 201620,China)
出处
《东华大学学报（自然科学版）》
CAS
北大核心
2023年第2期92-98,共7页
Journal of Donghua University(Natural Science)
基金
国家重点研发计划资助项目(2018YFB1308800)。
关键词
运动鞋
鞋帮
打磨轨迹
非刚性点云配准
贝叶斯相干点漂移
athletic shoes
upper
grinding track
non-rigid point cloud registration
Bayesian formulation of coherent point drift
