Frame filtering and path verification for improving video copy detection

Recently, the frame fusion based video copy detection scheme provides a possibility to detect copies in a continuous query video stream. However, its computational complexity is high since a large amount of returned reference frames need to be handled by some reference clip reconstruction methods. In addition, dense frame sampling strategies generally used for improving copy localization precision not only further aggravates the computational efficiency but also leads to much more false alarms due to content redundancy among frames. To alleviate the above problems, a new scheme is proposed for improving the performance of the frame fusion based video copy detection in both efficiency and effectiveness. In particular, the continuous similarity property among neighbor frames is learned for guiding the design of smart frame filtering method so as to greatly reduce the redundancy among frames. Then, an effective path verification scheme, which utilizes cross-clip verification strategy, is given for removing false alarms. The extensive experimental results show that the proposed schemes remarkably improve the detection accuracy of the baseline frame fusion scheme and give a comparable localization accuracy to it.