二维矢量地图双重零水印算法

摘要

为解决传统水印算法鲁棒性和不可感知性之间的矛盾,提出一种二维矢量地图双重零水印算法.该算法以矢量地图线图元特征点提取为基础,针对反映线图元基本形状的特征点,采用BP神经网络来构建特征点之间的关系模型,并由此构造能够抵抗简化、压缩攻击的零水印参数.针对反映线图元细节的非特征点,利用矩阵奇异值分解(SVD)方法来构造对常见几何攻击稳健的零水印参数.该方案很好地刻画了矢量地图线图元的整体特性,双重水印参数的构造有助于提高水印算法的抗攻击能力,达到对矢量地图全面保护的目的.实验结果表明,该双重零水印方案具有较好的鲁棒性和不可见性,适用于二维矢量地图的版权保护.%To address the contradiction between robustness and imperceptibility in a traditional watermarking algorithm, a double zero-watermarking algorithm for 2D vector maps was proposed. Based on feature point extraction,for the feature point reflecting the outline of the polyline, the algorithm utilized the feature point relation model established by a BP artificial neural network to construct a zero-watermarking parameter that was robust against simplification and compression attacks. For the non-feature point reflecting the detail of the polyline, the algorithm utilized the matrix singular value decomposition (SVD) to construct a parameter of zero-watermarking that was robust against geometry attacks. This algorithm perfectly depicts the whole character of each polyline, and the construction of double zero-watermarking parameters can improve robustness and achieve a comprehensive protection target for vector maps. Experimental results show that the proposed zero-watermarking has preferable robustness and invisibility, which is suitable for copyright protection of 2D vector maps.