Title :
Temporal Video Registration for Watermark Detection
Author :
Chupeau, Bertrand ; Oisel, Lionel ; Jouet, Pierrick
Author_Institution :
Thomson R&D France
Abstract :
Video frame alignment with the original source signal is required in a variety of applications, for instance non-blind watermark detection. Robustness to temporal cropping, scene removal, frame-rate changes and other non-linear temporal distortions are required. This paper presents an algorithm for locating a video segment within a long video document with frame-accurate alignment. The video sequences are first summarized into one-dimensional "temporal profile" signals. Then frame-to-frame correspondence is obtained by a dynamic programming approach. An automatic assessment of the matching result is proposed. Extensive test results achieved 77% success rate
Keywords :
dynamic programming; image registration; image segmentation; image sequences; video coding; watermarking; dynamic programming; frame-accurate alignment; frame-to-frame correspondence; nonlinear temporal distortions; one-dimensional temporal profile signals; scene removal; source signal; temporal cropping; temporal video registration; video frame alignment; video segment; video sequences; watermark detection; Cameras; Dynamic programming; Hidden Markov models; Layout; Nonlinear distortion; Research and development; Robustness; Signal processing; Video sequences; Watermarking;
Conference_Titel :
Acoustics, Speech and Signal Processing, 2006. ICASSP 2006 Proceedings. 2006 IEEE International Conference on
Conference_Location :
Toulouse
Print_ISBN :
1-4244-0469-X
DOI :
10.1109/ICASSP.2006.1660303
