Shot Boundary Detection at TRECVID 2007

Abstract

Shot boundary detection is one of the most funda-mental processes in video analysis, and it requires high detection accuracy and high-speed processing. This paper proposes a method of shot boundary detection based on multiple features. The pro-posed method enables precise, high-speed detection by omitting the processing of frames that are clearly not shot boundaries, and by analyzing various fea-tures only for the parts of the video that are likely to contain shot boundaries. An experiment using TRECVID 2007 test data resulted in a recall rate of 90.5 % and a precision rate of 94.4%. About 425 min-utes of test data was processed in 3 minutes 28 sec-onds (excluding the MPEG1 decoding time), 1/123 of the real time. 1