Title: Similar video detection using multiple direct-mapped cache

Authors: Noboru Murabayashi; Kenichi Yoshida

Addresses: The Research Group of Information Science and Social Science Fusion, University of Tsukuba, Ootuka 3-29-1, Bunkyo-ku, Tokyo 112-0012, Japan ' The Research Group of Information Science and Social Science Fusion, University of Tsukuba, Ootuka 3-29-1, Bunkyo-ku, Tokyo 112-0012, Japan

Abstract: Recently, the importance of video detection technology has increased. In our earlier studies, we had proposed the (single) direct-mapped cache method based on the RGB colour space feature to handle the processing of a large amount of videos on the internet. Although our method achieved a high processing speed, the detection accuracy had room for improvement. In this paper, we propose a new method to improve the accuracy. For this purpose, the combination of MPEG-7 video signature and the multiple direct-mapped cache method is used. The experimental results show that the average detection accuracy is 95% and the processing time is 0.2 seconds for each video. These results indicate a 29% improvement in accuracy over the previous method. The possibility of handling 136 years worth of video data while keeping the same accuracy and processing speed is also one of the unique features of the proposed method.

Keywords: similar videos; direct-mapped cache; MPEG7 video signature; LSH; hash; video detection; internet; detection accuracy.

DOI: 10.1504/IJISTA.2013.056522

International Journal of Intelligent Systems Technologies and Applications, 2013 Vol.12 No.3/4, pp.178 - 193

Received: 26 Nov 2012
Accepted: 31 Jan 2013

Published online: 25 Sep 2013 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article