Research Of Uncompressed Digital Video Watermark Technology Based On Wavelet Transform

With the development of multimedia and Internet technology, there are moreand more digital products for our entertainment and they can easily be copied anddelivered on Internet without any losses in quality. Therefore, more and morepirates strongly hurt the producer and author's interests .It calls for an effectivemethod to protect digital products .Previously, people have used encryption toresolve the problem, which can make digital products safe, however, it cannotprotect the content of digital products because the technology application isfinite .Digital watermark is a new technology to protect digital products becauseit is embedded in the multimedia data. The watermark may be all the copyrightinformation such as numbers, words and icons, which make it possible to protectcopyright, communicate secretly and inspect the truth of digital products.Digital watermark is some labels or signatures embedded into digitalmultimedia host data by means of signal processing, exploiting the characteristicsof Human Visual System (HVS) or Human Auditory System (HAS) fully. Thelabels or signatures usually cause unnoticeable perceptual distortion to theoriginal data and are retrieved only by an appropriate extraction algorithm.Digital watermark is such an effective method that it becomes the focus onmultimedia security research. Therefore, digital watermark and its keytechnologies have aroused great attention in academy and industry. Among all kinds of digital watermark technologies, researchers have paidmore attention to video watermark techniques due to its high research value andpotential market demand. But presently, video watermark techniques availablelag far behind against still image watermark techniques for the lack of accuratemotion vision model and the limitations of existing video coding standards andmeantime, some special attacks to video watermark techniques emerge gradually.In this paper, for the compressed videos, a video watermark algorithm basedon the MPEG-2 standard is proposed. Two kinds of video watermark algorithmsin the area of uncompressed video are introduced. Video watermark is embeddedand extracted based on DCT and DWT . Experimental results show that videowatermark is embedded and extracted based on DWT is better than others onrobustness of watermark.For the compressed video, we choose to insert the watermark informationwhich should be modulated by m sequence in the I frame chrominance of DCcoefficients because DC coefficients are steady parameters and exist in the videostream .The watermark information is deleted difficultly and the algorithm isrobust and it can stand against attack effectively. Chrominance of DC coefficientsare sensitive parameters for vision system, therefore, we embed watermark intoDC coefficients, which should be equal to going into minim interference .Wemust make the interference under the threshold, which can make human visionsystem sense immune from the changing. The watermark based on DCcoefficients is embedded, which can improve the digital watermark check-upquality and video effect. This method only needs portion decoding and it has alow computing complexity. During inserting, portion decoding and the DCcoefficient is used. By the embedding algorithm, the watermark is embedded intothe coefficient and the new coefficients are obtained;and then portion encoding isadopted and the watermarked compressed video is formed. While extracting,portion decoding and DC coefficient are obtained and through the extractionalgorithm, the watermark is extracted;whether the watermark is judged existusing the threshold.In the area of uncompressed video, we propose a video watermark algorithmbased on DCT .First of all, one frame will be changed from RGB color space toYUV, and then DCT is done on Y component, and taxis is implemented afresh fortranspositional coefficients at Zig-Zag scan mode;secondly, real numberpseudo-random sequences are embedded which meets N(0,1) as watermark intothe decomposed coefficients ,and make IDCT to coefficients which should beembedded on watermark, and then color space transformation is done from YUVto RGB, obtaining the watermarked frame;thirdly, the operation to every framein this video stream is the same. The watermark extraction algorithm is utilized toget the watermark and compare with the primary watermark and compute thecorrelation, which can judge whether the watermark is existent by the value ofcorrelation.The digital video watermark algorithm based on digital video scenesegmentation and 3D wavelet transform is presented in this paper .The algorithmis also a blind video watermark method which need not original video. For anyhost video, the video shot segmentation algorithm has been presented to partitionthe video into video shots with different scene. Then some key shots are selectedto embed watermark. For each selected video shot, the watermark information isembedded into 3D wavelet coefficient, i.e. the watermark information isembedded into each components of video sequence, including the static and thedynamic components. Watermark image for the protection of copyright sign canmake encryption pretreatment through a random shuffle matrix .In addition, thespread spectrum technique is adopted to generate a sequence of watermarkimages during the process of watermark embedding, which can make watermarkimage information become watermark image sequence adaptively because of thecharacteristic of video data and watermark image. Thus, the existence ofwatermark information redundancy will be effective to watermark detection.Finally, experimental results show that video watermark is embedded andextracted based on DWT is better than video watermark is embedded andextracted based on DCT and video watermark algorithm based on the MPEG-2standard on robustness and transparence of watermark.