Abstract
An incommensurable amount of digital audiovisual information is becoming available in digital archives, on the World Wide Web, in broadcast data streams and in personal and professional databases, and this amount is only growing. Mining short video repeats is just providing an effective way to extract useful information from fast growing video data and exploit the reuse value of video archives. Short video clips, ranging from a few seconds to a few minutes, are widely used in many types of video production, including news video, sports video, education program where they serve as program logo, station logo, sports reply flying logo or as commercials. These short video clips are repeated many times in these video programs because of the nature of their usages in video production. Identifying these short video repeats based on their contents has great value in many media applications. First it can be used to monitor commercials and detect infringement of copyrighted content in broadcast videos or web videos (Agnihotri et al., 2003; Cheung et al., 2005; Lienhart et al., 1997; Snchez et al., 2002; Kashino et al., 2003; Yuan et al., 2004), and this content based video identification approach is an important complementary method to other media copyright protection techniques such as watermarking. Second, short video repeat mining is very useful in video structure analysis task. By detecting video repeats in unlabeled raw video data, we can discover correlation of different video parts and structural video elements used for syntactic segmentation purpose, thus video structure model can be effectively constructed and applied to video syntactical segmentation (Yang & Tian et al., 2007). Structure analysis by finding repeat objects has also been extensively employed in DNA data mining domain (Kurtz & Schleiermacher, 1999; Bao & Eddy, 2002). Video repeat mining also has many other applications, such as content summary, personalization as well as lossless video compression (Pua & Gauch, 2004). Video repeats are defined as those video clips having the same video contents, but low level signal distortions are allowed, such as noises, image quality reduction, frame size change et al. Video repeat mining problem can be classified into two categories: one problem is to identify known video repeats. In this case we have a model of a sample video in advance, then use this model to detect its repeated instances in video archives, which is treated as a pattern recognition problem; The second problem is to identify unknown video repeats. In this case we do not have prior knowledge about the video repeats including their frame O pe n A cc es s D at ab as e w w w .in te ch w eb .o rg
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have