Video copy detection based on ORB and ordinal measure feature.

To address digital video copyright protection issue, this paper proposed a method which was based on ORB local feature that was a binary feature descriptor and ordinal measure global feature for video copy detection. The method detected shot from video through comparing Bhattacharyya distance of the adjacent frames' gray histogram and selected the first frame in shot as a key-frame. Then it extracted the ordinal measure feature and ORB feature from key-frames respectively. When query video came, it used ordinal measure feature to get the first result which could remove some irrelevant videos and then obtained the video copy result through ORB feature. Experimental results show the effectiveness and usefulness of the proposed approach for video copy detection, both the accuracy and recall are above 80%. [ABSTRACT FROM AUTHOR]