 |
    |
Block Operator Context Scanning for Commercial Tracking
Ioannis Giannoukos1, Vassilis Vrachnakis1, Christos-Nikolaos Anagnostopoulos2, Ioannis Anagnostopoulos3, and Vassili Loumos1
1Electrical & Computer Engineering School, National Technical University of Athens, Greece
[email protected]
[email protected]
2Cultural Technology and Communication Dpt., University of the Aegean, Greece
[email protected]
3Computer Science and Biomedical Informatics Dpt., University of Central Greece, Greece
[email protected]
Abstract. The industry that designs and promotes advertising products in television channels is constantly growing. For effective market analysis and contract validation, various commercial tracker systems are employed. However, these systems mostly rely on heuristics and, since commercial broadcasting varies significantly, are often inaccurate. This paper proposes a commercial tracker system based on the Block Operator Context Scanning (Block - OCS) algorithm, which is both accurate and fast. The proposed method, similar to coarse-to-fine strategies, skips a large portion of the image sequences by focusing only on Regions of Interest. In this paper, a video matching algorithm is also proposed, which compares image sequences using time sliding windows of frames. Experimental results showed 100% accuracy and 50% speed increase compared to traditional block-based processing methods.
Keywords: Block-Operator Context Scanning, Commercial Tracking, Sliding Windows, Video Matching
LNAI 7297, p. 369 ff.
[email protected]
© Springer-Verlag Berlin Heidelberg 2012