In this section, we verify the effectiveness of SRLP-FS through a toy example. From table II, we can know AT&T database has 400 samples, 10304 features and 40 classes. Each class has 10 samples, which represents a person's 10 face images. We randomly select two classes from AT&T database and choose two samples from these two classes as the test samples. We apply SRLP-FS to the two test samples and select {1280, 2560, 3840, 5120, 6400, 7680, 8960, 10240} features respectively to observe the feature selected results. We draw the results in Fig. 2. In Fig. 2, we maintain the original values of the selected features and set the unselected features to white. Fig. 2. Tow test samples from AT&T face database with different number of selected features.
