General Regression and Representation Model for Face Recognition

Jianjun Qian, Jian Yang; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2013, pp. 166-172

Recently, the regularized coding-based classification method (e.g. SRC and CRC) shows a great potential for face recognition. However, most existing coding methods ignore the statistical information from the training data, which actually plays an important role in classification. To address this problem, we develop a general regression and representation model (GRR) for classification. GRR not only has advantages of CRC, but also introduces the prior information and the specific information to enhance the classification performance. In GRR, we combine the leave-one-out strategy with K Nearest Neighbors to learn the prior information from the training data. Meanwhile, the specific information is obtained by using the iterative algorithm to update the feature weights of the test sample. Finally, we classify the test sample based on the reconstruction error of each class. The proposed model is evaluated on public face image databases. And the experimental results demonstrate the advantages of GRR over state-of-the-art methods.