Face Verification Based on SVM and LBP
Hui Wang, Jian Wu · International Conference on Electric Information and Control Engineering · 2012
Adaboost algorithm has been prevalent for a while in the field of face detection because of its effectiveness and efficiency. However, it still generates some false positive faces, which brings much difficulty to the subsequent process, such as face recognition, etc. To address the problem, this paper proposes a new method to verify the face images outputted by the adaboost algorithm. Specifically, it first extracts Local Binary Pattern (LBP) features from the candidate face images and then classifies them into face or non-face by Support Vector Machine (SVM). Extensive experiments have shown that our algorithm can achieve precision higher than 98%.