您的位置：首页 > 论文页面
基于KinectFusion的人体坐姿三维几何重建
发表时间：2018-05-31 浏览量：1476 下载量：651
| 全部作者： | 吴俊，游雅利，谭力凡，陶武康 |
| 作者单位： | 湖南大学汽车车身先进设计制造国家重点实验室；湖南师范大学工程与设计学院 |
| 摘 要： | 人体坐姿三维几何数据是开发人体有限元模型及乘坐用具的基础数据。本文设计并实现了一种基于微软KinectFusion技术的三维人体坐姿重建系统。该系统硬件由第二代Kinect（Kinect v2）及人体坐姿支撑平台组成，结合微软KinectFusion及自定义点云处理算法，实现对人体点云数据的采集与处理。通过试验设计方法确定了Kinect的最佳扫描位置及转台转速。利用该系统对26名志愿者进行坐姿三维几何重建。试验结果表明，系统有效解决了坐姿扫描过程中的遮挡问题，获得的人体坐姿三维模型的平均误差为0.9 cm，精度较高，同时，系统具有成本低、操作简单等优点。人体坐姿三维几何快速重建方法的建立，将为人体坐姿数据的获取提供支撑。 |
| 关 键 词： | 计算机应用；人体建模；深度传感器；点云处理；人体坐姿 |
| Title： | 3D geometric reconstruction of human body sitting posture based on KinectFusion |
| Author： | WU Jun, YOU Yali, TAN Lifan, TAO Wukang |
| Organization： | State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body, Hunan University; College of Engineering and Design, Hunan Normal University |
| Abstract： | Three dimensional (3D) geometric data of human body sitting posture is the basic data of developing human body finite element model and riding tool. 3D human body sitting posture reconstruction system based on Microsoft KinectFusion technology is designed and implemented in this paper. The system hardware, which consists of Kinect v2 and supporting platform for human body sitting posture, is combined with Microsoft KinectFusion and custom point cloud processing algorithm to achieve point cloud data acquisition and processing on the human body. The optimal scanning position of Kinect and the rotary speed of the turret have been determined through experimental design method. 3D geometry reconstruction of 26 volunteers’ sitting postures is processed by using this system. The test results show that the system effectively solves the problem of occlusion in the process of sitting position scanning, the average error of human body sitting posture 3D model is 0.9 cm with high precision. Besides, the system has the advantages of low cost, simple operation and so on. The establishment of 3D geometric reconstruction method for human body sitting posture will provide support for human body sitting posture data acquisition. |
| Key words： | computer applications; human body reconstruction; Kinect sensor; point cloud processing; human body sitting posture |
| 发表期数： | 2018年5月第10期 |
| 引用格式： | 吴俊，游雅利，谭力凡，等. 基于KinectFusion的人体坐姿三维几何重建[J]. 中国科技论文在线精品论文，2018，11（10）：964-972. |
1
评论数 0请您登录
暂无评论
下载全文
多维论文