Face recognition by using overlapping block discriminative common vectors

Authors:
Xiaohui Wang
Affiliations:
Department of Computer Application and Technology, HanShan Normal University, Chaozhou, Guangdong Province, China
Venue:
IScIDE'12 Proceedings of the third Sino-foreign-interchange conference on Intelligent Science and Intelligent Data Engineering
Year:
2012

Citing 10
Cited 0

Application of the Karhunen-Loeve Procedure for the Characterization of Human Faces

IEEE Transactions on Pattern Analysis and Machine Intelligence
Face Recognition by Using Discriminative Common Vectors

ICPR '04 Proceedings of the Pattern Recognition, 17th International Conference on (ICPR'04) Volume 1 - Volume 01
Discriminative Common Vectors for Face Recognition

IEEE Transactions on Pattern Analysis and Machine Intelligence
The equivalence of two-dimensional PCA to line-based PCA

Pattern Recognition Letters
Improving the interest operator for face recognition

Expert Systems with Applications: An International Journal
One improvement to two-dimensional locality preserving projection method for use with face recognition

Neurocomputing
A feature extraction method for use with bimodal biometrics

Pattern Recognition
Discriminant common vectors versus neighbourhood components analysis and Laplacianfaces: A comparative study in small sample size problem

Image and Vision Computing
Short Communication: A novel local preserving projection scheme for use with face recognition

Expert Systems with Applications: An International Journal
LPP solution schemes for use with face recognition

Pattern Recognition

Quantified Score

Hi-index	0.00

Visualization

Abstract

Discriminative Common Vectors (DCV) has been widely used in face recognition. Previous literatures show that DCV can outperform PCA or LDA in classification accuracy of face images. In this paper, the author proposes a novel block DCV method, i.e. overlapping block DCV. This method first partitions every image into a number of blocks and views each block as a sample. Calculating the covariance matrix and solving its eigen values and eigenvectors are similar to PCA. Then the method chooses any sample from each class and projects it onto the null space to obtain the Common Vectors. DCV takes the Common Vectors as transform axes and exploits the transform axes to perform feature extraction. Compared with conventional DCV, overlapping block DCV seems to be more robust to the variation of facial details such as facial expression and can obtain a higher classification accuracy for face recognition.