三维网格模型盲提取水印算法

Blind watermarking algorithm for three-dimensional mesh model

基于离散傅里叶变换(DFT)技术和扩频通信技术提出三维模型稳健水印嵌入算法。算法利用扩频通信技术将原始的一维二值水印序列生成可直接嵌入的水印信息;利用主成分分析(PCA)方法,将原始模型变换到仿射(旋转、平移和均匀缩放)不变空间中,选取模型顶点到其中心距离作为水印嵌入单元,采用单极性量化嵌入单元离散傅里叶变换系数幅度的方法实现水印的盲提取。对算法进行了仿真,结果表明嵌入的水印信息具有不可见性,且能够抵抗模型的旋转、平移、均匀缩放等常见攻击。

A novel watermark embedding algorithm for three-dimensional （3D） mesh model based on discrete Fourier transform （DFT） and spread-spectrum communication technique is proposed. The original one-dimensional （1D） binary watermark sequence is mapped to watermark information, which can be directly embedded into host, with spread-spectrum communication technique. With principal component analysis （PCA） method, a transformation of the model to invariant space is proposed prior to watermark embedding. The lengths between the vertices and the centroid of the model are chosen as the watermark embedding primitives. The watermark information is embedded into the host through modifying the amplitudes of the discrete Fourier transform （DFT） coefficients of the embedding primitives using polar quantization technique to achieve blind watermark extraction. The experiment results show the embedded watermark is invisible and withstands the affine transform （rotation, translation and uniform scaling） attacks.