3D Human Motion Retrieval Based on ISOMAP Dimension Reduction

In recent years, with the development and increasingly mature of motion capture technology, it has become one of the most widely used technologies to obtain realistic human motion in computer animation. With the increasing demands, motion dataset is becoming larger and larger. Due to motion feature data have the high-dimensional complexity, we first adopt nonlinear ISOMAP manifold learning algorithm to resolve the "curse of dimensionality" problem for motion feature data. In order to save the time of reducing dimension, we adopt the scarcity of neighboring-graph to improve ISOMAP algorithm for making it apply the massive human motion database. Then we build a motion string index for database, deploy Smith-Waterman algorithm to compare the retrieval samples' motion string with motion strings of candidate datasets, finally, we obtain the similar motion sequence. Experiment results show that the approach proposed in this paper is effective and efficient.

[1]  Tido Röder,et al.  Efficient content-based retrieval of motion capture data , 2005, SIGGRAPH 2005.

[2]  B. Prabhakaran,et al.  Hierarchical Indexing Structure for 3D Human Motions , 2007, MMM.

[3]  Yi Lin,et al.  Efficient Motion Search in Large Motion Capture Databases , 2006, ISVC.

[4]  Zhaoqi Wang,et al.  Efficient motion data indexing and retrieval with local similarity measure of motion strings , 2009, The Visual Computer.

[5]  Lizhuang Ma,et al.  Content-Based Human Motion Retrieval with Automatic Transition , 2006, Computer Graphics International.

[6]  Paisarn Muneesawang,et al.  Advances in Multimedia Information Processing - PCM 2009, 10th Pacific Rim Conference on Multimedia, Bangkok, Thailand, December 15-18, 2009 Proceedings , 2009, PCM.

[7]  Kiyoharu Aizawa,et al.  Content-Based Cross Search for Human Motion Data using Time-Varying Mesh and Motion Capture Data , 2007, 2007 IEEE International Conference on Multimedia and Expo.

[8]  B. Prabhakaran,et al.  Indexing 3-D Human Motion Repositories for Content-Based Retrieval , 2009, IEEE Transactions on Information Technology in Biomedicine.

[9]  Kozaburo Hachimura,et al.  Segmentation of Dancing Movement by Extracting Features from Motion Capture Data , 2008 .

[10]  Hans-Peter Seidel,et al.  Advances in Computer Graphics, 24th Computer Graphics International Conference, CGI 2006, Hangzhou, China, June 26-28, 2006, Proceedings , 2006, CGI.

[11]  Woong Choi,et al.  A Quick Filtering for Similarity Queries in Motion Capture Databases , 2009, PCM.

[12]  Wu Fei,et al.  Motion Retrieval Based on Large-Scale 3D Human Motion Database by Double-Reference Index , 2008 .

[13]  Zhigang Deng,et al.  Compression of Human Motion Capture Data Using Motion Pattern Indexing , 2009, Comput. Graph. Forum.

[14]  Zhigang Deng,et al.  Perceptually consistent example-based human motion retrieval , 2009, I3D '09.

[15]  Matthieu Guillaumin,et al.  Combining Image-Level and Segment-Level Models for Automatic Annotation , 2012, MMM.