人体散乱点云数据的区域分割算法

Research on Region Segmentation Algorithm of Human Scattered Point Cloud Data

正确的人体点云数据分析不仅是人体3D测量的必要手段,更是未来服装数字化设计的基础,也是服装定制化智能生产的数据来源。人体散乱点云数据相比规整数据拥有更多的噪声及不规则性,这使得提取轮廓以及提取分割特征点更加困难。为了解决人体散乱点云数据分割难题,提出了基于移动最小二乘的切割算法。首先使用主成分分析法进行点云数据的调整,并使用夹角分析法提取投影到特定平面的二维轮廓。在此基础上,采用移动最小二乘法对部分二维数据点进行局部拟合并根据导数信息提取分割特征点。最后,利用VT K作为点云显示平台,对不同人体点云数据进行算法验证。实验结果表明,该分割方法实用可靠。

Correct human body segmentation is not only a necessary means for 3D human body measurement,but also the basis for future digital design of clothing.It is also the data source of intelligent production of clothing customization.Scattered body point cloud data has more noise and irregularity than regular data,which makes it more difficult to extract contours and extract segmentation feature points.In order to solve the problem of data segmentation of scattered human point cloud,this paper proposes a cutting algorithm based on moving least squares algorithm.Firstly,the principal component analysis method is used to adjust the point cloud data,and the angle analysis method is used to extract two-dimensional contour projected to a specific plane.On this basis,moving least square method is used to extract the segment feature points according to derivative information.Using VTK as a point cloud display platform,algorithm verification is performed on different human point cloud data.Experimental results show that the segmentation method is practical and reliable.