W-系统矩和Fourier变换下Volume Descriptors不变特征的三维模型检索

3D Model Retrieval Based on W-Systems and Volume Descriptors Invariance of Fourier Transform

摘要: 针对V-系统矩难以构造快速算法的不足,提出一种W-系统矩及其快速算法,并融合volume descriptors不变特征应用于三维模型检索.首先对模型体素化;其次计算体素模型的W-系统矩和物体运动不变量,得到特征向量;最后进行特征融合及匹配.实验结果表明,该算法的检索效率和性能均优于三维球面调和函数方法和三维Legendre矩方法.

Abstract: In order to speed up the computational efficiency of V-system moment, a novel method is proposed based on W-system moment and volume descriptors (VD) invariance using Fourier transform for 3D model retrieval.Firstly, the models are voxelized.Then, W-system moment features and VD invariance features are extracted from voxelized models.Finally, feature fusion and matching are conducted.Theoretical and experimental results validate that the effectiveness of the proposed descriptor has high precision rate compare to 3D Legendre moment and spherical harmonic method.