135 Effective Content Based Video Retrieval System Based on Query Clip
Download citation file:
- Ris (Zotero)
- Reference Manager
In the recent years, digital image and video have emerged as the new techniques for the creation, distribution and storage of information. Latest developments in multimedia technologies facilitate the capture and storage of video data with comparatively low-cost computers. Moreover, the novel possibilities provided by the information highways have resulted in the widespread availability of large amount of video data. Nevertheless these data are highly futile unless appropriate search methodologies are employed. Contemporary users are dissatisfied with the video retrieval systems that offer analogue VCR functionality. Content-based search and retrieval of video data has turned out to be a demanding and significant issue. Video encompasses multiple types of audio and visual information those are hard to extract, combine or trade-off in common video information retrieval. In this paper, we address the specific aspect of inferring our approach for content-based video retrieval from a collection of videos. In particular, we introduce a new video data model that supports the integrated use of different approaches. Initially the system splits a video into a sequence of elementary shots, extracts a few representative frames from each shot and computes frame descriptors based on color, Edge and motion features. The extracted features of all the elementary video shots are stored in feature library. In our system, the videos are retrieved based on a query clip. For a query video clip, the color, edge and motion features are extracted and compared against the features in the feature library. The comparison is performed with the aid of Kullback- Leibler distance similarity measure. Afterwards, similar videos are retrieved from the collection of videos based on the calculated Kullback- Leibler distance.