A Novel Feature Fusion Approach Based on Blocking and Its Application in Image Recognition
According to the idea of canonical correlation analysis, a block-based method for feature extraction is proposed. The main process can be explained as follows: extract two groups of feature vectors from different blocks which belong to the same pattern; and then extract their canonical correlation features to form more effective discriminant vectors for recognition. To test this new approach, the experiment is performed on ORL face database and it shows that the recognition rate is higher than that of algorithm adopting single feature.
KeywordsFeature Vector Recognition Accuracy Canonical Correlation Analysis Canonical Variate Feature Fusion
Unable to display preview. Download preview PDF.
- 2.Liang, Y.X., Gong, W.G., Pan, Y.J., Li, W.H., Liu, J.M., Zhang, H.M.: Singular Value Decomposition-Based Approach For Face Recognition. Optics and Precision Engnineering 12(5), 543–549 (2004)Google Scholar
- 3.Weenink, D.: Canonical Correlation Analysis. In: Institute of Phonetic Science, University of Amsterdam. Proceedings, vol. 25, pp. 81–99 (2003)Google Scholar
- 5.Lattin, J.M., Douglas Carroll, J., Green, P.E.: Analyzing Multivariate Data, pp. 313–345. China Machine Press, BeijingGoogle Scholar
- 6.Hotelling, H.: Relations Between two Sets of Variates. Biometrika 8, 321–377 (1936)Google Scholar
- 7.Huang, D.S.: Systematic Theory of Neural Networks for Pattern Recognition. Publishing House of Electronic Industry of China, Beijing (1996)Google Scholar