Face Tracking Algorithm Based On Feature Matching

With the development of science and technology,and at the same time with the improvement of information technology.Face tracking research in various fields such as intelligent monitoring,interpersonal interaction,military and other fields gradually highlights its important application value.Face tracking technology which involves many disciplines,it has become the research hot spots of image processing,computer vision and pattern recognition.Not only in theory but also in applications has important significance.However,when in the process of tracking,facing the impact of illumination changes,occludes and the similar objects in cluttered environment.So in this paper,analyzing the results of previous research,and put forward the problems of face tracking,and at the same time find a way to solve the problem of it.The main research is how to find the adaptive feature model of the target,based on face tracking algorithm,so needing to match the target with the model to prevent the loss of tracking when the occlusion is happen.The main work of our paper can be summarized as follows:Firstly,this paper introduces the face tracking technology of the related research and the present situations,and simple description of the different tracking algorithms and the related application;then introduces the face tracking algorithm related technologies and research,and analysis of the specific steps in CAMSHIFT algorithm and basic principle of target tracking,then describe the facial feature description method;finally,the problem of occlusion were related,and puts forward some measures to solve the occlusion problem.Secondly,introducing the SURF algorithm.It is proposed to extract facial feature points to track object.In order to solve the problem of the CAMSHIFT algorithm which can not solve the loss tracking problem when occur the skin-color occlusion in the tracking process,put forward with the feature matching to track the object,and using the Speed Up Robust Feature to find the feature points which has the special nature that can reflect the image information in the video,and through the Fast Library for Approximate Nearest Neighbors matching method to match the right goals,with the obtained position and direction information of the feature points to constraint the CAMSHIFT algorithm.So it can ensure the stability of the tracking object.Thirdly,in order to more comprehensive description of the global and local feature information,put forward a face tracking algorithm based on AAM inversesynthesis algorithm.AAM is based on the shape model with combination of surface texture information of the global face region,and mainly introduce the principle of the algorithm and the method of the key feature points positioning,and the inverse synthesis algorithm which applied to the AAM.Finally,do some experiments to verify the feasibility and effectiveness of the algorithm,and proved through the establishment of the feature points extraction model can locate the target accurately,and improve the tracking of the anti-jamming capability.