摘要
三维人体测量技术中,人体表面情况较为复杂以及存在视觉盲点,使得测量范围有限的Kinect技术无法通过一次测量完成对整个人体表面点云的获取工作。为了得到与原模型一致的分块点云数据和解决Kinect直接获取的点云坐标与实际不相符的问题,研究了基于Kinect对人体的不同区域的测量,首先推导出合理的坐标变换公式对点云进行处理。然后,使用三点对齐法和Delaunay三角剖分法实现分块点云的拼接和三维人体模型的重建。最后,运用Matlab平台进行仿真。实验结果表明,该算法简化了传统三维人体建模的复杂性,能够精确地提取人体深度图像,且模型的恢复程度较好。
In three-dimensional body measurement technology, since the body surface is quite complex and has blind spots, it impossible for Kinect technology with limited measuring range to complete the acquisition work on the point cloud of full body surface by measuring only once. In order to obtain the blocking point cloud data being consistent with original model and to solve the problem that the point cloud coordinates directly acquired by Kinect mismatch with the actual ones, we studied the Kinect-based measurement on different parts of human body. First we deduced reasonable coordinate transformation formulas to process the point cloud data. Then we used three-point alignment method and Delaunay triangulation method to realise the splicing of blocking point cloud and the rebuilding of 3D human body model. Finally, we employed MATLAB to simulate the algorithm. Experimental results showed that the algorithm simplified the complexity of traditional 3D human body modelling, could accurately extract the deep image of human body, and was good in recovery of the model as well.
出处
《计算机应用与软件》
CSCD
2016年第1期219-221,248,共4页
Computer Applications and Software
