Blind 3D Mesh Watermarking Based On The Featured Areas

The current watermarking technology focuses on media types like still images, and video and audio streams. In contrast, the problem of watermarking 3D models has received less attention from researchers. As more and more CAD-based 3D data is entering the World Wide Web, companies or copyright owners who present or sell their products in virtual space will face copyright-related problems. The straightforward demand is to prevent their 3D-based material from unauthorized use. As a new way for the protection of information copyright, digital watermarking will possess extensive application perspective in this area.In 1998, Ohbuchi introduced three methods to insert the watermark into the mesh. In 1999, Praun introduced the algorithm to watermark after remesh, registration and other preparation. It is very important in the watermark research of the mesh. In domestic, Kangkang Yin of Zhejiang University introduced the method based on the multi-resolution processing. The algorithm is a frame of the watermark. It can be used into the other watermark methods of the mesh to increase the robustness, which has great significance. In 2007, Patrice Rondao Alface and his fellows firstly introduced the protrusion function into the watermarking scheme to get the feature points in their paper. And they defined the areas to be embedded watermark based on the feature points. But the approach is implemented based on the geodesic distance. The geodesic distance really can well reflect the feature of the mesh, while it increases much more complexity of computing. And the neighbor prongs also can influence the watermarking areas sometimes. This means that it is less robust against the cropping. Jae-Won Cho presents a new watermarking method in his paper released in 2005. This method divides the margin between the maximum and the minimum of the sphere radius into several equal parts. The number of the bins is equal with the number of the watermarking bits. After computing the mean of the radius of every bin, they compare the mean with the middle of the bin. At last, they modified the radiuses based on the comparing results and the bits to be watermarked. This method gives the inspiration for the blind watermarking scheme. But because of a bit determined by several sphere radiuses, the watermarking capacity becomes lower. In a word, although we have had a lot of good work, the watermark for the mesh still needs to be researched, especially in the frequency domain and which is blind.Because of the transforming of the watermark in the frequency domain and the complexity of the mesh watermark algorithm, it makes the efficiency of the mesh watermark become the problem which needs to be researched. The paper aims to find a new efficient method to watermark by reducing the computing with the high capacity. The difference between the frequency watermark and the geometry watermark is the different objects to be watermarked. The first kind of the watermark is always transforming the mesh by DCT, FFT, DWT and so on. Then watermark is inserted by modifying the frequency coefficients. The watermark in the frequency domain makes the change spread all over the mesh, so it is always more robust than the watermark in the geometry domain; the other kind is inserted by modifying the coefficients like the geometry, topology and the other property of the mesh。Based on the new research, the paper introduces a new blind mesh watermarking scheme in the frequency domain. First, we choose the prongs of the mesh which best reflect the feature of the mesh; secondly, to suppose that the prongs are the center of the circle and the length we choose is the radius, we can get the area which could be watermarked; and then ,with the sphere coordinate, we will get the distances from the object center to the vertices, and perform a DFT on it; the watermark will be embedded by changing the DFT frequency coefficients. At last, the watermarked 3D object is produced by reverse DFT and the coordinate transform from sphere to Cartesian. The method could make the mesh become robust to the cropping. The Discrete Fourier Transform can achieve that the embedded watermark is invisible and the scheme is robust against addition of noise. Experimental results show that this watermarking scheme doesn't make the quality of the mesh drop heavily and the method is robust against attacks such as rotation, translation, uniform scaling and even crop.