Research On Robust Video Watermarking Technology Against Geometric Attacks

With the rapid progress in copyright protecting and integrity authentication of multimedia, digital watermarking technology has been proposed as an effective method to protect the authenticity of intellectual property. And due to the characteristics of video-multimedia, the video watermarking has being one of the hotsports and difficulties in digital watermarking research. Some scholars have put forward many video watermarking schemes according to international video coding standards. At prensent, most watermarking algorithms proposed mainly address the robustness to common signal processing while neglecting geometric attacks. Therefore, research on robust video watermarking technology against geometric attacks take very important accademic and practical values.This thesis focuses on video watermarking algorithms resisting to geometric attacks. The main work is concluded as following several aspects.First, presents a zero-watermark algorithm for video which is based on the Krawtchouk Moment Invariants of original images to counterattack RST attacks. Due to the stability of the seven lower Krawtchouk Moment Invariants exhibited in geometrical transformation, the seven moment invariants of the video frames should be calculated first and then construct the watermark sequence with zero-watermarking technology. Throughout this algorithm, no watermarking information is embedded in the images and thus a good imperceptibility can be achieved. During the process of detection, no original video sequence is needed. The experimental results demonstrate that the algorithm is robust against RST attacks, cropping attack and other conventional attacks.Second, Rotation attack can destroy the synchronization of the watermark detector and the watermarking, even a slight rotating may invalidate the detection of watermarks. Resisting to the rotation attack, a blind video watermarking algorithm is proposed on the basis of combinding the Zernike moments and singular value decomposition (SVD). First choose the embedding areas on the luminance components. Then make the partitioning while doing the discrete wavelets transform (DWT) and make the singular value decomposition to the low-frequency parts. And last embed watermarks through modifying the maximum singular value. In detection, the possible rotation attacks to the video sequence would be corrected through Zernike moments. The experimental results show that the proposed algorithm is robust against conventional attacks and arbitrary angle of rotation attacks.Third, a video watermarking scheme based on geometric moment corrected is proposed. This algorithm used the geometric moment's good characteristic of corrected parameters of scaling attack, and combined with the rendomness and fuzziness of Cloud model. In watermarking forming process we uses the Cloud model. In detection process, first corrected the detected video sequence's parameters of possible scaling attack by using geometric moment, then extracted watermark from corrected video sequence. The experimental results show the robustness to lossy compression, filtering, frame dropping, additive noise, scaling and translation attacks.