Cascade boosting LBP feature based classifiers for face recognition

Local Binary Pattern (LBP) is a powerful means of texture description that has achieved great success in face analysis area. In this paper, we propose a face recognition approach using boosted LBP-feature based classifiers.The multi-class problem of face recognition is transformed into a two-class one of intra- and extra-class by classifying every pair of face image as intra-class or extra-class ones. The cascade framework, is used to overcome the problem of overwhelmingly large number of samples and grossly imbalance of the positive and negative samples. By boot-strapping negative examples, sub-training spaces (random subsets) are randomly generated, and then weak classifiers are learned using every sub-training space (random subset). The weak classifiers are combined into a strong one by improving recognition accuracy. Experimental results on FERET database show competitive performance.