摘要
提出一种基于曲面参数化理论的半自动匹配方法.首先,选择任一样本数据作为参照模型,根据其表面几何特征手动标定特征标记点;其次,利用保角映射方法将全部样本数据映射到同一球形域,根据球面坐标对应关系实现在其他样本表面上特征标记点的自动标定;最后,利用迭代逼近点方法归一化特征标记点在空间的位置.实验考察了统计形状模型的通用性原则和专一性原则,并且比较了球面保角映射中不同约束条件对获得模型质量的影响.实验结果表明:利用球面保角映射3个基准点约束所得模型的专一性指标明显好于零质心约束方法,并且实验所需时间减少近50%.
A semi-automatic matching method based on surface parameterization theory is presented. Firstly, one of surface sample data as a reference model is selected. According to the geometric characteristics of surface, some landmark points are choosed manually. Then, all the sample data are mapped to the same sphere by the conformal mapping method. The landmark points on the other surface samples are completed automatically according to the positions corresponding to their spherical coordinates. Finally, to normalize the positions of landmarks in threedimensional space, the Iterative Closed Points method is utilized. The experiments measure the general principle and specificity principle of the model. Moreover, the influence on the point distribution model due to the different constraints of spherical conformal mapping method is compared. The results show that the specificity of the model based on spherical confomcal mapping theory is much letter than that of zero centroid constraint method, the time for the experiment is decreased by 50%.
出处
《天津工业大学学报》
CAS
北大核心
2016年第1期54-58,共5页
Journal of Tiangong University
基金
天津市应用基础与前沿技术研究计划项目(14JCYBJC42300)
关键词
球面保角映射理论
特征标记点匹配
统计形状模型构建
活动形状模型
图像分割
spherical conformal mapping theory
signature point matching
statistical shape model building
active shape model
image segmentation