Star: Semi-supervised-Clustering Technique with Application for Retrieval of Video

Semantic video retrieval is a key application in today´s networked world. With the increasing proliferation of digital video contents, efficient techniques for analysis, indexing and retrieval of videos according to their contents have become ever more important. This paper proposes the new method of STAR (Semi-supervised clustering Technique with Application for Retrieval of video) in the web mining application using the data mining algorithms. This Process is performed by the content based image retrieval and other distance functions. Conventional distance functions are Mahalanobis Distance, Euclidean Distance, and Bregman Distance are used to evaluate the distance between Point-to-Point and Point-to-Centroid. The combined effect of MKBoost and Bregman K-means clustering algorithm will give promising results in the retrieval of images from video structural semantics. Experimental results show that the proposed algorithm is more effective and efficient than the existing algorithm.