Face recognition based on rearranged modular 2DPCA
ICIC'11 Proceedings of the 7th international conference on Advanced Intelligent Computing Theories and Applications: with aspects of artificial intelligence
Probabilistic learning of similarity measures for tensor PCA
Pattern Recognition Letters
| Hi-index | 0.00 |
Recently, in a task of face recognition, some researchers proposed a novel method called Extended Two-Dimensional PCA (E2DPCA) which is an extension to the original 2DPCA for gray face recognition. This method can get a new covariance matrix preserving more local geometric structure information than previous 2DPCA methods and avoid small sample size problem of PCA method. But this approach only considers the global information of face images and only is applied for gray face recognition. Some local information and color information may be ignored. Therefore, in this paper, the sub-pattern technique was combined with E2DPCA based on quaternion matrix for color face recognition. In this paper, this new sub-pattern based E2DPCA(which is called SpE2DPCA) was performed and compared with PCA, 2DPCA, E2DPCA, Sub-pattern PCA(SpPCA) and Sub-pattern 2DPCA(Sp2DPCA) for face recognition. The numerous experiments show that the proposed this novel approach SpE2DPCA based quaternion matrix has better recognition performance than other traditional PCA-based methods for face recognition.