高级检索
邹歆仪, 李桂清, 尹梦晓, 柳雨新, 王宇攀. 变形图驱动和变形感知的谱姿态迁移[J]. 计算机辅助设计与图形学学报, 2021, 33(8): 1234-1245. DOI: 10.3724/SP.J.1089.2021.18667
引用本文: 邹歆仪, 李桂清, 尹梦晓, 柳雨新, 王宇攀. 变形图驱动和变形感知的谱姿态迁移[J]. 计算机辅助设计与图形学学报, 2021, 33(8): 1234-1245. DOI: 10.3724/SP.J.1089.2021.18667
Zou Xinyi, Li Guiqing, Yin Mengxiao, Liu Yuxin, Wang Yupan. Deformation-Graph-Driven and Deformation Aware Spectral Pose Transfer[J]. Journal of Computer-Aided Design & Computer Graphics, 2021, 33(8): 1234-1245. DOI: 10.3724/SP.J.1089.2021.18667
Citation: Zou Xinyi, Li Guiqing, Yin Mengxiao, Liu Yuxin, Wang Yupan. Deformation-Graph-Driven and Deformation Aware Spectral Pose Transfer[J]. Journal of Computer-Aided Design & Computer Graphics, 2021, 33(8): 1234-1245. DOI: 10.3724/SP.J.1089.2021.18667

变形图驱动和变形感知的谱姿态迁移

Deformation-Graph-Driven and Deformation Aware Spectral Pose Transfer

  • 摘要: 为了提高姿态迁移过程中网格表面形状细节的保持能力,减少分层姿态迁移的交互环节,提出变形图驱动且变形感知的自动分层谱姿态迁移方法.首先利用变形图对三维模型进行形状保持的全局低频姿态迁移;然后根据模型变形前后的特征变化自动分割出局部刚性块,并对它再次进行姿态迁移,直到所有局部网格的姿态得到充分迁移.通过多个例子对文中方法和Yin等方法进行了比较,实验结果表明,文中方法次级姿态迁移的次数降低38.0%,平均距离误差降低54.0%,表面积和体积的变化降低12.5%.该方法的姿态迁移较充分,在模型的形状保持上更有优势,且自动化程度较高.

     

    Abstract: A deformation graph driven and deformation aware spectral pose transfer method is proposed in order to improve the ability of the spectral pose transfer method in preserving shape details and reducing user interac-tions.First,a deformation graph is introduced to drive the source model deforming under the guidance of low-frequency spectral coefficients of the reference model to obtain a deformed source model.Second,the de-formed source model is automatically segmented to parts according to the deformation.Third,repeating the spec-tral pose transfer on two source parts separately yields two deformed source parts.This process can be repeated again if necessary.Finally assembling all deformed parts in a reversed order results in the completely deformed source model.Comparison with the method of Yin et al.is conducted via a couple of examples.Experimental results show that the secondary pose migrations are reduced by 38%,the average distance error is reduced by 54%,and both surface area and volume variations are reduced by about 12.5%.It improves the quality of transfer results to a certain extent,reduces the number of middle-frequency pose transfer,and also reduces manual interaction.

     

/

返回文章
返回