dc.description.abstract | In recent years, 3D graphic models have become more accessible to general users due to the convenient use of advanced scanning devices and the virtual-reality modeling language (VRML) for graphic description. Moreover, due to the growth of Internet, many issues of 3D data processing such as 3D visualization, transmission, compression, and security problems gain more and more attention. In this dissertation, we propose the methods on two interesting 3D research issues: 3D fragile watermarking and 3D progressive compression.
There are two major purposes for 3D fragile watermarking: integrity checking and changed region locating. Two problems frequently arise in the embedding stage: the causality problem and the convergence problem. The causality problem arises while the neighboring relationship of a former processed vertex is influenced by the perturbation of its latter processed neighboring vertices. The convergence problem means that the original model has been heavily distorted before some vertices reach the predefined relationship. In this dissertation, we propose a method to overcome these two problems. The proposed method is a public scheme which means that it does not need the original model and watermarks for authentication. The key for extracting watermarks is relatively smaller than that of the previous works. Our method can control the average distortion by the keys used in watermark embedding. Based on this technique, we also developed a transformation invariant fragile watermarking technique for 3D model authentication. The main contribution of the proposed scheme is that it is invariant to translation, rotation, and uniformly scaling operations. We think these operations do not change the integrity of the original models and should not be regarded as a specific forgery. The proposed technique holds all advantages that provided in the prior scheme and provides an extra transformation invariance characteristic. From the viewpoint of technique, these two schemes are quite different.
Moreover, a geometry-driven hierarchical compression technique for triangle meshes is proposed such that the compressed 3D models can be efficiently transmitted in a multi-resolution manner. In 3D progressive compression, we usually simplify the finest 3D model to the coarsest mesh vertex by vertex and thus the original model can be reconstructed from the coarsest mesh by operating vertex-split operations in the inversed vertex simplification order. In general, the cost for the vertex-split operations will be increased as the mesh grows. In this dissertation, we propose a hierarchical compression scheme to keep the cost of the vertex-split operations being independent to the size of the mesh. Most previous 3D progression compression schemes first encode the connectivity information, and then predict and encode the geometry information based on the connectivity information. We propose the geometry-driven technique that predicts and encode the connectivity relationship of vertices based on their geometry information. Experimental results show the effectiveness of the proposed encoding scheme. | en_US |