Abstract: Topological features detection and automatic segmentation of the point clouds from human body surface in standard standing measuring posture are studied，and a new method based on panoramic range image is proposed which is used for topological features detection and automatic segmentation of the point clouds from human body surface.Point clouds from human body surface are first converted into cylindrical polar coordinates and the panoramic range image of human body is created.Several key landmarks on human surface are detected automatively according to the layer distribution of the panoramic range image.Human scans are segmented into five functional parts in terms of the human key landmarks detected.The experimental results show that the method can improve the efficiency and accuracy of the topological features detection and automatic segmentation of the point clouds from human body surface in standard standing measuring posture.

摘要： 对标准站立测量姿态下的人体表面点云数据的拓扑特征检测与自动分割进行了研究，提出基于全景深度图像表示的人体点云表面拓扑特征检测和自动分割新方法。首先把人体表面的点云数据转换为圆柱极坐标形式，获得人体扫描表面的全景深度图像表示，根据全景深度图像中的层次信息自动检测人体表面的拓扑特征，并根据拓扑特征把人体分割成5个功能结构。实验证明这种方法改进了人体表面点云数据的拓扑特征检测和自动分割的效率和精度。