Video Face Clustering And Character Relationship Analysis

With the gradual change of peopel's audio-visual consumption in the information age,the number of TV,movies and other video resources increases rapidly.Meanwhile,with the continuous development of video analysis technology and the demand for automatic access to the information such as video story line,characters growing,video face clustering and character relationship analysis,as a kind of technology to excavate the potential information in video,has become a hot research.In this paper,based on the analysis of the state-of-the-art of video face clustering and video character relationship analysis,we further study the following research and gain some achievements.Different from the traditional image-based face clustering method,the video face clustering algorithm will deal with some situations such as the changes in brightness,angle and resolution,occlusion,and expression changes in face images,and there are a lot of faces in the video,which poses a very high demand for the effectiveness and scalability of the clustering algorithm.Based on the above problems,this paper proposes a video face clustering algorithm based on pairwise constraints.The algorithm uses the temporal and spatial information of the face Tracks in the video to obtain the priori information,and then uses the information to get the pairwise constraints within the faces,which are introduced into the hierarchical clustering algorithm to enhance the clustering effect.At the same time,in order to enhance the effectiveness and scalability of the algorithm in dealing with the large scale of face data in video,we use the nearest neighbor chain and data partition optimization algorithm to reduce the time complexity of the algorithm.To verify the validity of the proposed method,we conducted experiments on a real movie videos,experimental results fully reflect the effectiveness and feasibility of the method.Based on the temporal and spatial context and video semantic information of the scene,this paper proposes a method of exploiting character co-occurrence relationship,and constructs character's social network called SRN through the quantitative relationship among the characters.Based on SRN network,we can get rid of the limitation of the traditional feature-based approach and carry out more in-depth video semantic analysis.By analyzing the characters in the SRN network,a community identification method based on the core characters is proposed,which can automatically confirm the core characters in the video and dig out the community around the core characters.In this paper,lots of movie video are used to experiment,and experimental results show the effectiveness of SRN model and community identification method.