No abstract

-The traditional kernel two-dimensional principal component analysis (K2DPCA) method did not take full advantage of the class information for face images and there are both "outer class" problem and "hard classifier" problem on face recognition. Therefore, a new face recognition method based on fuzzy kernel two-dimensional principal component analysis (FK2DPCA) is presented . Firstly, it introduces fuzzy concept into K2DPCA. Secondly, the class separability of criterion will be extended to high dimensional feature space by the use of kernel method. Furthermore, we select the eigenvectors that betweenclass scatter is greater than within-class scatter after projection as optimal projection axis. Finally, it uses the nearest neighbor classifier for face recognition . The experiment results on ORL and YALE face databases show that the FK2DPCA is better than other traditional methods.

To measure the similarity between two images, the local information around all the points were usually used for traditional image retrieval methods based on interest points. The accuracy would be influenced by the dissimilar interest points in the background region or in the uninterested regions. A new image retrieval method based on interest points was proposed in this paper. The most similar interest points were chosen firstly based on feature matching techniques, then the similarity was measured based on the local gray information around the preserved similar interest points. Experiments show that the proposed method is more accurate than traditional image retrieval method based on interest points.

No abstract