A novel SVM classification approach in tensor-faces algorithm
HU Ai-rong, Shan Jiang · 2010
Multi-view face recognition is still an important and challenging problem to face recognition. In this paper, we propose an improved approach basing on Tensorfaces algorithm which focuses on how to improve the feature extraction and the classification methods to make the recognition accurately. SVM is a classifier that has demonstrated higher generalization capabilities in many pattern recognition problems. The SVM Classifier is used in the proposed method instead of Nearest Neighbor Classifier in the Tensorfaces algorithm. The proposed method is evaluated on the Weizmann face image database. Experimental results show the performance of the method is better than the original TensorFaces method.