Face Detection using Variance based Haar-Like feature and SVM

摘要: This paper proposes a new approach to perform the problem of real-time face detection. The proposed method combines primitive Haar-Like feature and variance value construct feature, so-called Variance based feature. Face in image can be represented with small quantity features using this We used SVM instead AdaBoost for training classification. made database containing 5,000 samples 10,000 non-face extracted from real images learning purposed. contain many which have differences light conditions. And experiments showed that detection system much more efficient than AdaBoost. tested our on two databases one Non-Face database. obtained 96.17% correct rate YaleB database, is higher 4.21%