摘要
针对传统矢量地图零水印算法存在的抗几何攻击和要素增删攻击能力不足等问题,该文提出一种基于ResNet50模型的矢量地图零水印算法,在不改变矢量地图数据的情况下保护地图版权。该算法通过采用道格拉斯—普克算法提取矢量地图的复杂特征信息,将坐标集合映射为适合ResNet50模型处理的序列,通过ResNet50模型自动提取矢量地图数据中的特征,以实现零水印的构建。选择西安市餐饮、村庄、乡镇村道矢量地图数据集作为实验数据进行算法验证,结果表明,该算法在抗几何攻击和要素增删攻击方面均表现出色,在无损版权保护方面具有广阔的应用前景。
Aiming at the problems of insufficient ability to resist geometric attacks and element addition and deletion attacks in traditional vector map zero watermarking algorithms,this paper proposes a novel zero-watermarking algorithm for vector maps based on convolutional neural networks(CNNs),aiming for lossless copyright protection of vector map data.The algorithm employs the Douglas-Peucker algorithm to extract the feature elements of vector maps,maps coordinate sets into sequences suitable for the ResNet50 model processing,and then automatically extracts features from the vector map data to construct the zero watermark through the ResNet50 model.Selecting the vector map datasets of catering,villages,and township roads in Xi′an as experimental data for algorithm verification,the experimental results shows that the proposed algorithm exhibits excellent performance in resisting both geometric attacks and element addition/deletion attacks,effectively protecting the copyright of vector map data.It also holds broad application potential in lossless copyright protection for vector maps and contributes to safeguarding the commercial interests of data providers.
作者
张永利
卢浩
ZHANG Yongli;LU Hao(Guangdong Surveying and Mapping Product Quality Supervision and Inspection Center,Guangzhou 510075,China)
出处
《地理与地理信息科学》
CSCD
北大核心
2024年第6期1-5,共5页
Geography and Geo-Information Science
基金
广东省十四五基础测绘专项项目。