Research And Implement Of Multi-View Face Detection Based On Real AdaBoost Algorithm

As a key technology in face information processing, face detection attracts a widespread attention in the field of computer vision and pattern recognition in recent years because of its high value of academic research and commercial applications. With the development of intelligent information processing technology, face detection makes a lot of applications in the identification, video encoding, content-based retrieval, automatic control, human-computer interaction and so on.This paper summaries and analysis the current typical algorithm of face detection, and then improves and optimizes the process of training and face detection. This paper realizes the multi-view face detection system and real-time face detection system on this basis. Firstly, this paper puts forward the idea that limits the height, width and area of the Haar features. This method decreases the number of features and training time. Secondly, for the huge data of training, this paper proposes the method which is to batch processing by writing files. Thirdly, this paper proposes a new method to find negative samples. It takes false positive which gets by scaling the detection window when detecting nonface image as negative samples. When computing Haar feature value, it divides by magnification. Finally, in real-time face detection system, this paper uses the DirectShow technology, which has significantly increased system speed to 18 frames per second.Our program about face detection has a good portability and lays a good foundation for future research and application.