Kullback-Leibler similarity measures for effective content based video retrieval

Abstract

Recent advancements in the multimedia technologies allow the capture and storage of video data with relatively inexpensive computers. As the necessity to query these data competently becomes significant, the amount of broadly accessible video data grows. As a result, content-based retrieval of video data turns out to be a demanding and vital problem. In this paper, an effective content-based video retrieval system is proposed. The raw video data are segmented into shots and the object feature, movement feature and the occlusion feature are extracted from these shots and the feature library is utilised for the storage process of these features. Subsequently, the Kullback–Leibler distance is computed among the features of the feature library and the features of the query clip which is extracted in the similar manner. Hence, with the aid of the Kullback–Leibler distance, the similar videos are extracted from the collection of videos based on the given query video clip in an effective manner.

Keywords:

• video retrieval
• content-based video retrieval (CBVR)
• video sequence
• shot segmentation
• object feature
• movement feature
• occlusion feature
• query clip
• similarity measure
• Kullback–Leibler distance