A novel robust video fingerprinting-watermarking hybrid scheme based on visual secret sharing

Abstract

In this paper, a novel robust Video Fingerprinting-Watermarking Hybrid Scheme (VFWHS) is proposed by applying fingerprinting technique to zero-watermarking algorithm for enhancing the performance of video Digital Rights Management (DRM). In the VFWHS, the content-based feature of the original video is extracted by an improved 3D-DCT fingerprinting algorithm and stored in a video fingerprint database for retrieval, while the watermark-based feature of the watermark image is extracted by a 2D-DCT fingerprinting algorithm for copyright identification. According to the visual secret sharing scheme, a master share is generated from the content-based feature while an ownership share is generated from the master share and the watermark-based feature. The ownership share is then stored in an ownership database. When a video is queried, its content-based features are generated, and a process of similarity measurement is executed to obtain the matched feature with the relevant ownership share. The copyright ownership can be identified by stacking the master share generated from the queried video with the relevant ownership share. Our experimental results demonstrate that the VFWHS can obtain the relevant ownership share precisely for reliable copyright identification in the mass content processing which cannot be achieved by the zero-watermarking algorithms, and provide explicit copyright identification to avoid possible copyright dissension which cannot be achieved by the fingerprinting algorithms.