Face Recognition with Single Training Sample Based on Local Feature Fusion
Chen Yang, Shuicai Shi, Lin Li, Xueqiang Lv · 2011
In the condition of single training sample, traditional methods get low recognition accuracy, or even can not be used. In view of this situation, this paper proposes a method to slove this problem. Firstly, face image is decomposed by image pyramids. Then, each layer image segmentation into sub images with th same size. After that, the feature of each sub image, which got with (W2DPC)2A, gets a weight through the adaptive method. Finally, Euclidean distance is used to classify face images. Experimental results on ORL and Yale show that the presented method can achieve a certain degree of recognition accuracy.