关节特征约束的3维模型骨架提取算法

Articulation constrained skeleton extraction of 3D models

根据多分辨率Reeb图（MRG）的原理，提出一种基于关节特征约束的骨架优化算法。它克服了基于曲率约束提取骨架方法中逐点计算顶点的曲率约束轮廓的低效性，通过分析网格顶点的离散高斯曲率，获取模型表面上具有凹陷特性的双曲极值点作为约束点，进行关节特征区域的有效提取。进而增加关节特性点，优化MRG骨架。实验结果表明，本方法有效地突出了模型的拓扑分支特征以及模型表面的细节，提高了骨架提取的精度和效率。

We propose an optimized articulation-constrained skeleton extraction approach based on multi-resolution Reeb graph （MRG）method. It overcomes the low efficiency of calculating the discrete contour constrictions of each vertex on a 3D model. Instead, it obtains the hyperbolic extreme points by analyzing the Ganssian curvature of each vertex and defines them as constrained points which imply the concave feature of the local surface, and the articulation feature areas are then extracted. Finally the Reeb graph is optimized by adding new feature points, and the constrained skeleton keeps topological feature and local details of 3D modes well. A series of experimental results have shown its accuracy and efficiency.