Co-segmentation of 3D shapes via multi-view spectral clustering
- PDF / 1,059,367 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 64 Downloads / 241 Views
O R I G I N A L A RT I C L E
Co-segmentation of 3D shapes via multi-view spectral clustering Pei Luo · Zhuangzhi Wu · Chunhe Xia · Lu Feng · Teng Ma
Published online: 10 May 2013 © Springer-Verlag Berlin Heidelberg 2013
Abstract Co-segmentation of 3D shapes in the same category is an intensive topic in computer graphics. In this paper, we present an unsupervised method to segment a set of meshes into corresponding parts in a consistent manner. Given the over-segmented patches as input, the cosegmentation result is generated by grouping them. In contrast to the previous method, we formulate the problem as a multi-view spectral clustering task by co-training a set of affinity matrices derived from different shape descriptors. For each shape descriptor, the affinity matrix is constructed via combining low-rankness and sparse representation. The integration of multiple features makes our method tolerate the large geometry and topology variations among the 3D meshes in a set. Moreover, the low-rank and sparse representation can capture not only the global structure but also the local relationship, which demonstrate robust to outliers. The experimental results show that our approach successfully segments each category in the benchmark dataset into corresponding parts and generates more reliable results compared with the state-of-the-art.
P. Luo · Z. Wu () · C. Xia · L. Feng · T. Ma Department of Computer Science, Beihang University, XueYuan Road No. 37, HaiDian District, Beijing, China e-mail: [email protected] P. Luo e-mail: [email protected] C. Xia e-mail: [email protected] L. Feng e-mail: [email protected] T. Ma e-mail: [email protected]
Keywords Co-segmentation · Multi-view clustering · Sparsity · Low-rankness
1 Introduction Co-segmentation of shapes is a crucial task for shape analysis and processing, which segments a set of 3D models into meaningful parts and creates their correspondence simultaneously. Such a segmentation can be utilized in various applications, such as modeling [1], labeling [2], and shape retrieval [3]. Recently, various methods have been proposed to segment a set of shapes consistently. In regard of the requirement of manually labeled data, these methods can be categorized into unsupervised [4–6], semi-supervised [7] and supervised methods [2]. In this paper, we propose an unsupervised method, where the co-segmentation problem is formulated as a multi-view spectral clustering task by cotraining a set of affinity matrices derived from different shape descriptors. In previous methods, the set of 3D shapes to be segmented consistently is commonly from the same category. Although the semantic parts inherent in these shapes are consistent, there exist large variations among them in geometry and topology. Therefore, it is insufficient to achieve satisfactory results using only one kind of shape descriptors. In order to improve the co-segmentation quality, more shape descriptors are beneficial. In the supervised method [2], features are selected by JointBoost, which requires a training dataset. Hu
Data Loading...
