A Method of Constructing Vertebral 3D Statistical Model Based on Gaussian Curvature
Aiming at the difficult problem of inaccurate model of medical spine 3D statistical model library, this paper studies a method of constructing medical spine 3D statistical model based on the feature points of Gaussian curvature flow localization model.In this method, the human lumbar vertebrae model is physically positioned based on the feature points of the Gaussian curvature flow to generate the sample matrix of feature points of each spine sample model.Then the sample matrix of the feature point is aligned and registered by the ICP iterative algorithm.Finally, PCA analysis is used to train and study the spine model samples after registration, and finally a more accurate vertebral 3D statistical model library is obtained.By comparing the performance parameters of the experimental results, the model constructed by this method is more accurate than before.
3D model ICP Gaussian curvature PCA
Jing DU Bin YU Yu HUI Jun-Sheng WU Chen ZHANG
School of Management Northwestern Polytechnical University, Xian, China College of Computer Science XI DIAN University Xian, China College of Computer Science Northwestern Polytechnical University, Xian, China College of Software and Microelectronics Northwestern Polytechnical University Xian, China
国际会议
成都
英文
101-106
2018-03-12（万方平台首次上网日期，不代表论文的发表时间）