Study On System Of Face Detecting And Tracking In Video

In the dissertation, we carried on the study of Face Detection and Face Tracking. Face Detection and Face Tracking technology is the key technology in the computer vision. Facial image information processing of the Computer vision includes face recognition, profile estimating, facial expression recognition, video monitoring etc., and almost all of those are related to face detection and face tracking. Learning from the domestic and international discourse and research papers on face detection and recognition in recent years, firstly a systematic summary of previous work is given in this paper. An automatic face tracking detection system is developed here, which is on the basis of the previous research on face detection and tracking. The system achieved crucial improvements on some traditional and classical algorithms of face detection and facial feature location. My primary work is as following:Two face detection methods are proposed in our prototype system. One is using Adaboost learning algorithm and the other is based on skin color model in HSV chrominance space. The common characteristic of the algorithms is the rapid detection speed and at the same the two methods are complementary.Using face detection method based on Adaboost learning algorithm, which selects few key haar-like features from a large set of features, to build a robust cascade classifier. A rapid face detection system is designed and realized on the basis of the extension. The system can detect frontal images including a wide range of formats. The detection experiment with the CMU databases shows that the system reached a high hit-rate and low false-alarm-rate. Especially, the detecting speed of the system is very high and almost attains the criterion of real-time.In skin color model face detection, a color transformation technique is applied to the face images in the HSV chrominance space. Then the skin regions could be segmented by a series of threshold to the skin tone model. On considering to clearing the skin-similar noise in the back-ground, an algorithm using the difference between frames is applied, which has been proved effective by practice. After the face areas are obtained, a method to get their centers and sizes quickly is proposed here and then face-areas would be signed.Kalman algorithm is used in the paper for face tracking, which combines both the face detection and the object tracking. Face detection algorithm is used to determine the initial tracking regional: after obtaining face objects by Adaboost algorithm, skin color detecting algorithm is used here to check the detecting veracity. This could improve the detecting precision. The Kalman algorithm is applied for tracking with the objects'information after they are ascertained. Kalman algorithm forecasts the objects'information in the next frame. Then matches are done to make sure whether the tracking is right or not. If right, the objects records should be renewed and the results are signed. Kalman algorithm is simple calculation, fast tracking.