Research On Watermarking Algorithms For NURBS Model Based On Wavelet Transform

With the rapid progress of the technologies for digital media and Internet,the copyright Protection of digital media has been an important research field. Current watermarking technology focuses on still images,video and audio streams. But,more and more 3D digital media have been used in CAD/CAM,the straight forward demands are to prevent the 3D NURBS model from being used unauthorized. As a new way of the information copyright protection, digital watermarking will process extensive application perspective in this field.This dissertation first introduces the related backgrounds of watermarking technologier of 3D geometry model, its framework and the classification, and its research progress. Then, the dissertation systematically studies the basic theory of the NURBS curves and surfaces and the basic theory of wavelet transform, and analyzes the main problems existed for the NURBS modle watermarking.A watermarking algorithm based on wavelet transform is proposed to deal with NURBS models. Firstly, virtual images are gotten based on the control points of the NURBS representation of the original NURBS model, Subsequently, these virtual images are decomposed by wavelet decomposition and the watermarking information is embedded in the low frequency coefficients. Finally, the marked virtual images are gotten with inverse wavelet transform and a new group of control points is obtained based on the pixel value of the virtual image. In this procedure, the watermarking information is embedded and a marked NURBS model is gotten. Experiment shows that the proposed method has the feature of transparency and robustness against the operations of knot insertion/removal, knot refinement, degree elevation/reduction, and affine transformations.In addition, this dissertation proposes an algorithm of watermarking of NURBS models based on the non-uniform rational B-spline wavelets transform to fix the shortcomings of the previous algorithm. Firstly, the sub chain of the knot vector is elected for the needful LF(Low frequency) size. Subsequently, the control points make the decomposition of the non-uniform B-spline wavelets based on the sub chain. And then the watermark is quantization embedded in the LF part. Finally, the marked new control points are gotten with the reconstruction of the non-uniform B-spline wavelets. Experiment shows that the proposed method has the feature of transparency and robustness against the operation of knot insertion/removal, knot refinement, degree elevation/reduction, and affine transformations. At the same time, the algorithm is a blind extract algorithm to the quantization embedding.A series of digital watermarking simulation experiments and analyses are given for every algorithm in the dissertation. Some worthy conclusions are also gotten through analyzing the experimental results. The theory analyses and the experimental results show that the given algorithms are effective and they all have a good robustness and capacity against attacks.