Combining local and global features for face recognition (ヒューマン情報処理)

概要

論文の詳細を見る
In the literature of psychophysics and neurophysiology, many studies have shown that both global and local features are crucial for face recognition. In this paper, we summarize our work on face recognition by combining both local and global discriminative features. In our work, Fourier transform is exploited to extract global features from the whole face image domain. For local feature, Gabor wavelets and its combination with Local Binary Patterns (LBP) are explored, which are both conducted on some spatially partitioned image patches. These features are then fed into different classifiers based on Fisher Discriminant Analysis respectively, and these classifiers are combined together to make the final decision. The proposed methods are evaluated by using Face Recognition Grand Challenge (FRGC) experimental protocols and database, the largest data sets available. Experimental results on FRGC version 2.0 dataset show that our methods have much higher verification rates than the baseline of FRGC and the best known results under various situations such as illumination changes, expression changes, and time elapses.
社団法人電子情報通信学会の論文
2007-08-27