Near-duplicate keyframe (NDK) detection starts to attrack researchers attention. NDKs have an important role to reduce redundancies in query results, in copyright violations, commercial tracking and also news topic tracking. NDKs can also be used in Copy Detection, where copies of videos are detected in large databases, in Media Tracking, where a piece of media is tracked in different sources such as commercial tracking, and also in Story Tracking, where events that are captures by different cameras are tracked such as news tracking. However, existing approaches are limited by single frames only and discards sequential relations. Those approaches must be extended to sequences since digital media is growing day by day and analysing these media by shorter clips (sequences) is more meaningful. We mainly combine NDK detection with a tree-based sequence detection method where NDK detection is used to create rankings for frames. In this paper, we propose a new method to search different instances of a video sequence inside a long video. The proposed method is robust to view point and illumination changes which may occur since the sequences are captured in different times with different cameras, and to the differences in the order and the number of frames in the sequences which may occur due to editing. The algorithm does not require any query to be given for searching, and finds all repeating video sequences inside a long video in a fully automatic way. First, the frames in a video are ranked according to their similarity on the distribution of salient points and colour values. Then, a tree based approach is used to seek for the repetitions of a video sequence if there is any. These repeating sequences are pruned for more accurate results as a last step. Results are provided on two full length feature movies, Run Lola Run and Grounghog Day, on commercials of TRECVID 2004 news video corpus and on dataset created for CIVR Copy Detection Showcase 2007.

DATE: 24 August, 2007, Friday@ 10:00
PLACE: EA 409