Technology Of Video Authentication Based On 3-D Zernike Descriptors And Digital Watermarking

As an emerging technology that can be used for covert communication in the open network environment, digital watermarking has become a hot research topic of modem secret communication with its application in multimedia authentication. At present, research on multimedia authentication mainly focuses on image authentication. There is relatively little research conducted on video authentication, and there is no effective authentication scheme, especially the effective algorithm that can combine the inter-frame authentication with the intra-frame authentication.The works in this thesis are presented as follows:(1) The semi-fragile feature of 3-D Zernike descriptor in the video environment is verified. Simulation research is carried out by using different methods:adding noise of different strength, Mpeg2 compression with different bit rates, cutting of pieces of different sizes at different positions, replacement of different frames and selecting different video content from the same sequence. The results show that 3-D Zernike descriptor has small change under normal operations (such as noise adding and Mpeg2 compression), and it changes greatly under malicious operations (such as inter-frame attacks and intra-frame attacks). These results verify the semi-fragile feature of 3-D Zernike descriptor in the video operation. In addition, the semi-fragile feature of 3-D Zernike descriptor is studied by embedding watermark into video sequences. The result shows that the change of 3-D Zernike descriptor is similar to the change under normal operation. The change is equal to the change of adding noise in video sequences. This result further proves that 3-D Zernike descriptor can be used as characteristic quantity in video watermarking authentication.(2) A video authentication algorithm is proposed based on 3-D Zernike descriptor. In the algorithm, the Parametric Integer Wavelet Transform using the Lifting Scheme is applied for three-dimensional transform of the video sequence. Then the 3-D Zernike descriptor is extracted from low frequency coefficients and it is used as watermark embedded in low frequency coefficients. By using the approach of Structural embedding, the amount of computation is reduced while the embedding strength is enhanced. The result shows that the embedding of watermark has little influence on visual quality of images. Based on the semi-fragile feature of 3-D Zernike descriptor and the approach of Structural embedding, the algorithm can distinguish between normal operations and malicious attacks, and differentiate between inter-frame attacks and intra-frame attacks. Tampering would also be effectively positioned.This thesis makes contributes to the research of video watermarking authentication technology based on 3-D Zernike descriptor. The result proves that 3-D Zernike descriptor is reliable in video watermarking authentication. This paper contends that security of authentication algorithm would be strengthened with the application of Parametric Integer Wavelet Transform using the Lifting Scheme in video authentication. It also proposes a method based on 3-D Zernike descriptor and the approach of Structural embedding that can position the video contents which are tampered with. A series of experiments prove that this algorithm is efficient. This paper provides new methods for video watermarking authentication technology. It also provides new technologies to ensure the safety and reliability of video in application.