Video Classification Method Based On Clustering

With the rapid development of computer technology and Internet technology,a large number of multimedia data are produced on the mobile communication terminal and the Internet platform and other sources.Video is a kind of multimedia data that is rich on the semantics and quantity.On the basis of large quantity video,applications such as video indexing,video retrieval and video classification are produced.Video classification is a basic research topic in the field of computer vision,which is also an important intermediate of solving video annotation and video retrieval.In addition to those,the video classification is also the important method and means for multimedia data management.So the video classification is much valuable in research and applicationsVideo is a kind of multimedia data with rich semantic.For those videos with similar physical characteristics,they always have certain relevance.In this paper,on the basis of in-depth analysis of the structure of the video,video is segmented into shots,and then according to the similarity videos having the similar shots video classification is realized.In this paper the following research works are finished:(1)Some common video classifiers are deeply analyzed and some common video feature extraction methods are discussed.(2)An improved k-means clustering algorithm is put forward.In the algorithm the initial cluster centers of k-means are formed by the labeled sample,then videos are classified by features group which is formed by color features,texture features and SIFT features.The algorithm of this paper is compared with some common classification algorithms.Experiments show that the algorithm of this paper has higher classification accuracy and classification accuracy.The results of this paper can be widely used in applications such as video indexing,video retrieval and video classification,etc.