Video Face Detection Based On AdaBoost

Face detection is the prerequisite and basis of face recognition, crime analysis techniques, it has been widely used in the field of video surveillance, and it is a combination of pattern recognition, artificial intelligence, information security, and many other disciplines. At the same time, Face detection has a hidden, it can be monitored without fitting the case of target and provide effective protection for intelligent monitoring. This paper mainly studies face detection in the video, but face detection in the video need to consider many factors, such as the complex environments, multiple people face, face rotation angle, and so on. People need to find a face detection method with strong adaptation, high detection rate, and low false detection rate. But now the algorithm for the face detection can not fully adapt to all situations, so the establishment of the algorithm with a strong stable, the high detection rate is a challenging research topic.The study of video face detection is based on AdaBoost algorithm in this paper. The algorithm makes full use of high precision, fast detection speed's advantages and achieves real time face detection. The main work is done as follows:First of all, this paper analyzes the AdaBoost algorithm of face detection. This algorithm is divided into the following sections in accordance with the basic principles of AdaBoost algorithm: feature extraction section, classification training section and detection section. Feature extraction section details a method for selecting the features and quickly calculating Haar-Like features. The classification training section discusses the weak classifiers, the strong classifiers and cascade classifier. And this section details the step of training. The detection section discusses the principle of the multi-scale detection and the combined detection window. This section analyzes the detection results and influencing factors.In the second part, this paper proposes a new method for precise location of the human eyes based on AdaBoost. For interference problems of the glasses in the human eyes location, this paper fully exploits the eye structure information, combined with a fixed feature of glasses, proposes a new method for precise location of the human eyes. This method takes full advantage of the class structure of the eye round, and effectively avoids the impact of the glasses in the human eyes location.Finally, this paper proposes method of face detection in video, which combine AdaBoost algorithm and skin color segmentation. For interference problems of the complex environments and face rotation angle in the video face detection, this paper proposes a video face detection algorithm of double Gauss skin color model based on YCbCr. Experimental results show that this new color model has been low false alarm rate and missing rate, and it can better avoid the complicated background and color on the face detection interference.Experimental results show that this new color model has been low false alarm rate and missing rate, and it can better avoid interference of the complicated background and color for the face detection and have better face detection. This article is intended to provide a high detection rate, good stability, fast real-time face detection algorithm that provide theoretical support for improving monitor intelligent and the building of wisdom security.