In this paper, we proposed method to develop 3D skeleton and 3D object clustering. In 3D skeleton, Firstly,we use multi-view human images and find the feature points between difference angles by Speeded Up Robust Features (SURF) method. Second, we use an effective coordinate transformation method to transform feature points in 3D space. Third, we use improvement K-means algorithm, add three direction, to find the human join
points and to produce a simple 3D skeleton. In 3D object segmentation, we use Shape Diameter-Function (SDF)method and Gaussian Mixture Model (GMM) to segment regions in 3D model. In SDF method, we use SDF method to compute the SDF value by center of shape information and neighbor of current shape path information. In GMM method, we use GMM method to obtain the scope value of object clustering. Finally, we show results of our method in experiment results, and results show that our method is effective.
Relation:
Proceedings of 2011 Fourth International Conference on Ubi-media Computing, pp.168-173