3D Human Body Model Skeletonand Joints Extraction Based On Point Cloud Datafrom3D Scanning

With the advent and development of non-contact3D body scanning technology,We can getthe point cloud data of the whole human body in a relatively short period of time.But the pointcloud data with huge amount of information,lacking of necessary semantic information.Throughthe further data processing,we can get some important semantic information such as end futurepoint,topological structure,center line and the joint center etc.Because the center line can wellreflect the human body model skeleton,is a one-dimensional representation of3D models,so thecenter line is also known as the model of the skeleton.End of feature points and joint center arepart of the center line. Center line extraction is an important condition for the construction of realhuman body model animation.The extraction of skeleton and joint center is an importantresearch area in the field of computer graphics and visualization.In this paper,the triangular meshsurface reconstruction algorithm for generating Crust human body model of Voronoi map andDelaunay map based on the point cloud data obtained by3D scanning,basing on geodesicdistance of Morse function and Dijkstra algorithm to calculate the terminal feature pointsinformation,then get the information of human model head end feature point and feature pointsof extremities.Through the model of Morse function contour segmentation based on geodesicdistance, make a calculation of each layer of the geodesic distance contour model,obtaining thelevel-set-curve is the topology structure.Choosing the end feature point of head as the startingpoint,the center of gravity of the level-set-curve fitting adjacent formation center line so as toobtain the skeleton model. Take section circularity criterion to judge which level-set-curve thejoint center belong to. By calculating the center of gravity of the level-set-curve, so as toobtain model joint center.