| Summary: | 博士 === 國立中興大學 === 資訊科學與工程學系所 === 98 === Watermarking schemes have been widely used for protecting the authorization of digital content such as still images, video streams, audio streams, and three dimensional (3D) models. Among different media types, watermarking 3D models is comparatively hard inherent. Recently, a considerable progress has been made in the areas of steganography and robust watermarking on 3D polygonal meshes. Only a few fragile watermarking algorithms have been proposed to authenticate the integrity of 3D models. In addition, published three dimensional (3D) fragile watermarking schemes have many drawbacks, such as the causality, convergence, synchronization, and localization problems. Moreover, most fragile watermarking schemes for 3D models are not reversible. Based on these observations, this dissertation proposes reversible fragile watermarking schemes for three dimensional (3D) models in the spatial domain.
According to the purpose of integrity verification, capability of accurately verifying, locating, and recovering any tampered vertex is desirable in three dimensional (3D) fragile watermarking techniques. Furthermore, a fragile watermarking scheme is required to be robust against so-called content-preserving operations, including vertex reordering and similarity transformations. In this study, a principal component analysis (PCA) is first employed to translate the coordinates of the original points to the PCA-coordinate system and make the system robust against similarity transformation attacks. The points’ coordinates for each axis are then sorted to generate intervals. Moreover, using a secret key to embed the watermark insures security. Then two innovative schemes are introduced that embed fragile watermark into legitimate intervals of the 3D models in the spatial domain. The first irreversible fragile watermarking scheme is based on vertices perturbing to achieve that 100% vertices of the cover model can be embedded and extracted. A reversible fragile watermarking technique for 3D models in spatial domain is presented in the second scheme. PCA, together with a novel interval embedding technique, provide the fragility and blind reversibility. Intervals larger than a threshold and smaller than another threshold are adopted as legitimate to avoid large distortion.
In general, the steganography or watermarking schemes based on the synchronization technique have the problem of de-synchronization. The third reversible fragile watermarking scheme is proposed to overcome the synchronization problem. Unlike the method used in the first and second schemes, by modulating the distance from the vertex coordinate to the gravity center of the cover model, a reversible quantization index modulation embedding scheme is adopted to watermark each vertex of the cover model. And, the spread spectrum technique is adopted to improve the robustness while obeying the sensitive fragility and low distortion requirements.
In conclusion, this dissertation proposes one irreversible fragile watermarking scheme and two reversible fragile watermarking schemes for 3D models in the spatial domain. Experimental results show that these schemes can achieve high verification capacity with insignificant visual distortion of the original model. The causality, convergence, and localization problems are overcome in both the first and second schemes, and the synchronization problem is conquered in the third scheme. It is also shown that the proposed scheme is robust against vertex reordering and similarity transformation attacks. Finally, these results correspond to fragile watermarking requirements: security, reversibility, high capacity, low distortion, and accurately localization of modification.
|
|---|
