Efficient Sparse PCA-based Method for Motion Recognition
Jian Xin Xiang · Journal of Information and Computational Science · 2014
The emergence of substantial databases for capturing 3D human motions has become a new challenge to the efficient analysis and processing of human motion data. Such large data hinder the effective use of human motion capture databases. To reduce such high dimension complexity, sparse PCA is used to re-express motion data with the use of sparse representation. In particular, the projection from highdimensional data is decreased to reduce the dimensional subspace and to facilitate the recognition and classification of different motions. Consequently, automatic recognition and automatic retrieval of 3D human motions are realized.