An efficient approach for video retrieval by spatio-temporal features

Article type: Research Article

Authors: Naveen Kumar, G.S.^* | Reddy, V.S.K.

Affiliations: Faculty of Engineering, Lincoln University College, Malaysia

Correspondence: [*] Corresponding author: G.S. Naveen Kumar, Faculty of Engineering, Lincoln University College, Malaysia. E-mail: srinivasa@lincoln.edu.my.

Abstract: The rapid development of video capture and information sharing technology has resulted in an overwhelming number of online video archives. This makes it difficult to retrieve videos from a large database using a traditional text-based system. The only solution to this problem is to retrieve videos based on their content. Ample algorithms have been hypothesized to reclaim videos from an enormous data base. Besides they could not diminish the time consumption and their coherence couldn’t satisfy the users. We postulated the new approach which clubs the spatial features along with temporal features by making use of widespread video data and this amplifies the efficiency of video retrieval. In this regard, we move oncolour and motion features to get full data of video. Comparing the features of a demand video to those of a video to be retrieved from a server is easy, rather than comparing the entire content of the video. Therefore, designers of CBVR systems will follow two major steps to balance extraction and similarity of features [24].

Keywords: Video data base, feature extraction, video shot transitions, representative frame extraction, measurement of similarity, retrieval of videos

DOI: 10.3233/KES-190420

Journal: International Journal of Knowledge-based and Intelligent Engineering Systems, vol. 23, no. 4, pp. 311-316, 2019