RETRIEVAL OF VIDEO DATABASE USING R-SEGMENT TREE REPRESENTATION

Abstract

Rapid advances in storage and networking technology have enabled the support of video-on-demand applications using computer networks. As time goes on, the modes by which users can select the movie that they wish to watch are likely to increase dramatically. A multimedia system supporting user interaction must have the ability to efficiently retrieve the movies that satisfy the user's requirement. Furthermore, the user should have the option of watching a particular segment of the movie. In addition, the user should have the option of seeing critical reviews of the movie. Such reviews may help him to select desired movie for watching. For successful Multimedia database Management Systems (MMDBMS), it must have the ability to retrieve media object (i.e. video) from a local storage device in a smooth jitter free manner. As video data often occupy a large amount of space, therefore, new techniques are required for the fast retrieval and efficient use of storage. In the present work, a new approach for fast retrieval of any particular segment is developed. R-trees basically were used for image retrieval. We are extending this approach for video data representation under the name of R-segment tree representation. These databases are used for efficient retrieval of particular activity of any video. User can interact with the database through graphical user interface. Results show that proposed technique can retrieve desired activity of any video randomly with higher retrieval [iv] efficiency. The program for implementing this technique was developed in visual basic language.