Techniques for audiovisual data confidentiality, integrity and copyright protection

The fast growth of multimedia processing and delivery technologies makes many new multimedia applications possible. To prevent multimedia content from being stolen or illegally used, audiovisual content protection techniques become increasingly important so that the full potential of these applications can be reached. In this research, we investigate novel techniques for audiovisual data confidentiality, integrity and copyright protection, which are the most fundamental content security issues in the community.; We first propose two methodologies for fast encryption, i.e. selective encryption and integrated encryption and entropy coding based on multiple statistical models. The low computational cost is essential to multimedia confidentiality since the size of a multimedia data file is large and its real-time processing is often required. A selective encryption scheme for G.723.1 speech coding is presented, and two rules to determine whether a multimedia compression system is suitable for selective encryption are given. Furthermore, we propose a novel encryption framework by modifying entropy coders so that multiple statistical models are alternatively adopted in a secret order. Two encryption schemes are constructed by applying this methodology to Huffman coder and QM coder.; For speech content integrity, we investigate two fragile watermarking methods and a feature extraction scheme integrated with CELP speech coders. Both fragile watermarking schemes eliminate the participation of agents in the channel since no auxiliary data are required. However, the exponential-scale odd/even modulation is not compatible with CELP speech coders, and the linear additive watermarking scheme could fail to detect certain localized content alterations. The proposed feature extraction scheme is tolerant to many kinds of content preserving operations, including CELP coding.; Finally, a digital audio watermarking scheme of low complexity is proposed as an effective way to deter users from misusing or illegally distributing audio data. The proposed scheme is based on audio content analysis by using the wavelet filterbank while the watermark is embedded in the DFT domain. Special attention is paid to the synchronization attack caused by casual audio editing or malicious random cropping, which is a low-cost yet effective attack to previous watermarking methods.