传统水印嵌入方法往往导致原始载体数据质量的下降, 而可逆水印技术由于能完整恢复原始载体数据, 因此,在精度要求较高的矢量地图版权保护应用中具有较高的适用性。目前, 栅格图像的可逆水印算法研究较为成熟, 而在矢量地图领域则主要借鉴栅格图像的相关方法, 由于没有充分利用矢量数据的组织特性, 因此鲁棒性不强。针对上述问题, 本文以矢量地图坐标点之间的基本空间关系在常规地图操作前后的不变性, 提出一种鲁棒的矢量地图可逆水印算法。算法首先利用道格拉斯—普克方法将要素点分为特征点与非特征点两类, 在此基础上计算每个非特征点与其相邻前后特征点连线所构成的夹角。在水印嵌入时, 利用该夹角来映射水印比特位, 并通过调整非特征点坐标来对该夹角进行调制以实现水印信息的嵌入。实验表明该算法所生成的水印具有一般可逆水印的不可感知性与可逆性, 同时, 对旋转、平移、要素排序、裁剪、压缩、简化等攻击也具有较好的鲁棒性。

There is a decline in the quality of the original carrier when watermarking is embedded using the traditional technology. It can not meet the requirement of high-precision occasion and under the circumstance when original carrier data can not be tampered. Because the technology of reversible watermarking can completely recover the data of original vector map, it is highly suitable for the copyright protection with the high-precision requirement, for instance, military maps and maps provide evidence for the court. At present, the researches of reversible watermarking algorithm for image field become more mature, while the current reversible watermarking algorithm for vector map mainly refers to the relevant methods of the image field, having not fully considered the organization characteristic of the vector data, so the robustness is not strong. Aimed at the above issue, this paper carries out relevant researches on the algorithm of reversible watermarking of vector map. Based on the invariance of the basic spatial relationship of the points in vector map before and after conventional map operation, a robust reversible watermarking algorithm of vector map is introduced to resolve the problem. Firstly, Douglas- Poiker algorithm is used to classify the points of each feature into characteristic points and none characteristic points. Then the angle, constituted by the lines of each none characteristic point and two adjacent characteristic points, is computed by this algorithm. Finally, the watermarking bit is mapped by the angle and the water information is embedded into none characteristic point by modulating the angle mentioned above. The experiment shows that the watermarking using this paper’s algorithm not only has the transparency and reversibility of the general reversible watermarking, but also has good robustness to normal attacks, for instance, rotating, zooming, translating, feature ordering, graphic carving, data compressing, and graphic simplification, etc.