运用特征不变量的高精地图零水印算法

A Zero-Watermarking Algorithm for High-Precision Maps Based on Feature Invariant

高精地图作为车辆导航和自动驾驶的核心组成部分,具有精度高、包含敏感信息、成本高等特点,其安全和保密至关重要。数字水印作为信息安全领域的重要工具,可以为高精地图数据提供认证和保护,但目前缺少针对高精地图的水印算法。该文根据高精地图的特性,提出一种高精地图零水印算法:根据高精地图用于表示车道信息的参数三次曲线数量确定水印索引;利用两个车道信息节点中参数三次曲线数量的奇偶性是否相同确定水印信息;结合水印索引依照投票原则构建特征矩阵,与加密后的水印图像进行异或运算生成零水印。实验结果表明,该算法在受到较高程度的几何攻击、增删攻击后仍能提取到较完整的水印信息,具有良好的鲁棒性。

As a core component of vehicle navigation and autonomous driving,high-precision maps are characterized by high accuracy,sensitive information,and high cost.Ensuring their security and confidentiality is crucial.Digital watermarking,an import ant tool in the field of information security,can provide authentication and protection for data.However,due to the current lack of watermarking algorithms specifically designed for high-precision maps,these maps face a series of security issues during transmission and application.Aiming at the characteristics of high-precision maps,this paper proposes a zero-watermarking algorithm for high-precision maps.First,the watermarking index is determined based on the number of parametric cubic curves used to represent lane information in high-precision maps.Secondly,the watermarking information is determined based on the parity(odd or even)of the number of parametric cubic curves in two lane information nodes.Finally,combined with the watermarking index,a feature matrix is constructed according to the voting principle and the XOR operation is performed with the encrypted watermarking image to generate a zero-watermarking.Experimental results show that this algorithm can still extract relatively complete watermarking information after being subjected to high-level geometric attacks and addition and deletion attacks,demonstrating the good robustness.