A Study On OpenCV-based Algorithms For Face Detection And Tracking

Face detection and tracking technology, as an important component of the field of computer vision, can be applied in many aspects, such as social security, audio-visual entertainment, video conferencing, identification and so on. This thesis introduces the current status of the research and development of face detection and tracking at home and abroad. It gives a detailed analysis of the current algorithms of face detection and tracking nowadays and focuses on the improvement of the classical algorithm of it. The main contents are arranged as follows:First, on the basis of a detailed description and analysis of typical human face tracking algorithm and in accordance with the human skin color in the HSV color space clustering, this thesis proposes a new algorithm that mixes the color segmentation on HSV color space and Adaboost algorithm together. The experiment shows that, compared with Adaboost algorithm, which doesn’t use the skin color segmentation, the new algorithm can not only reduce the false detection rate but also increase the speed of face detection. It also proves that the increase of the detection speed is in inverse proportion to the size of the video.Second, on the basis of a detailed description and analysis of typical human face tracking algorithm, this thesis further studies and improves the classic Camshift algorithm as well as proposing a new algorithm by using the Kalman filter algorithm to correct Camshift algorithm when encountering interference. The experiment shows that, compared with Camshift algorithm, the improved algorithm can track and predict the target more accurately. Finally, the thesis builds the experimental platform to reach the algorithm implementation by utilizing OpenCV and Visual Studio2010, after which it goes on to analyze and compare the experimental results.