Semi-automatic registration of videos for improved watermark detection

Abstract

Virtually every video watermarking technology can benefit from comparison with the original content. For non-blind schemes it is fundamental; for others it is an improvement to increase the watermark's signal-to-noise ratio by subtracting the content that is often noise to the detector. A direct frame-by-frame comparison of the videos is not possible due to the fact that illegal copies of videos usually differ significantly from their originals caused by different spatial resolution or frame rates, geometric distortions from capturing, or targeted attacks. In this paper, we present a software tool that enables the semi-automatic temporal and spatial synchronization of frames and pixels of two similar videos. This process is called registration. We put our focus on utilizing human capabilities with the smallest possible effort, to allow a high overall performance and precision of the registration. An efficient graphical user interface supports the users and visualizes the results of all steps. In addition, we specifically distinguish digitally reproduced copies from analog (camcorded) copies in which two or more frames are blended into a new frame.

Rate-Distortion-Preserving Forensic Watermarking Using Quantization Parameter Variation

“…The original watermarked videos do not need to be stored in a large database, but can be generated on the fly using the original unwatermarked video. Although non-blindness may be considered a limitation, an advantage of such techniques is that they can assume temporal and spatial alignment [24], since existing video registration techniques can be used if they are not synchronized [34].…”

“…Since the proposed method is non-blind, this paper does not consider spatial or temporal synchronisation attacks such as rotation, framerate or camcording attacks. That is because existing video registration techniques can be used to synchronize the attacked video [34]. Future work could explore the effectiveness of these registration techniques when combined with the proposed method.…”

Algorithms for video retargeting

“…Examples are corner detectors like Harris [25] or SUSAN [75], or detectors that are invariant to image transformations like SIFT [54] or SURF [3]. The latter detectors are especially advantageous in case of artificial distortions of frames [69]. Frame rates of at least 25 fps are typical in videos that were produced for cinema or television, leading to relatively little camera motion between consecutive frames.…”

Robust digital watermarking in videos based on geometric transformations

“…Also, the geometric misalignment resulting from distortions has to be compensated (spatial synchronization) in order to be able to detect the transformations. Both are done using our video registration toolkit application and algorithms developed and presented in previous work [3].…”

“…The geometric synchronization is based on finding corresponding feature points as described in [3], and is performed once on each synchronization interval. For all frames of the following one or more encoding intervals, the temporally corresponding frames of the original, unmarked video are aligned to the distorted copy frames so that the only geometric differences should be those introduced by the watermark encoding.…”