基于AHP-模糊综合评价与BP神经网络的含水印矢量地图保真度评价方法

A Fidelity Evaluation Method for Watermarked Vector Maps Based on AHP-Fuzzy Comprehensive Evaluation and BP Neural Network

数字水印是矢量地图安全防护的前沿技术,但水印嵌入会影响地图质量。现有研究通常依赖视觉观测和误差分析判断含水印矢量地图的保真度或水印不可见性,忽略矢量地图本质特征,导致评价结果不精准。该文提出一种基于模糊神经网络与层次分析法的含水印矢量地图保真度评价方法,根据指标的不确定性和重要性构建含水印矢量地图保真度综合评价模型,综合考虑拓扑质量、几何特征和坐标误差等关键要素,通过模糊综合评价和BP神经网络的主客观权重优化,提高评价的准确性和客观性。该方法得出的含水印矢量地图保真度评价结果能更全面、准确地反映数据实际情况,较传统误差分析方法更科学合理;该方法可针对不同的应用场景和需求,根据不同数据集的特点调整水印算法的参数,为平衡水印算法的综合性能提供依据。

Digital watermarking,a cutting-edge technology for vector map security,can impact the quality of the maps to a certain extent.The existing research usually assesses watermarked vector maps′fidelity and invisibility via visual observation and error analysis,often ignoring the maps′intrinsic features,resulting in imprecise evaluations.Addressing these issues,this paper proposes a fidelity evaluation method for watermarked vector maps based on the analytic hierarchy process(AHP)fuzzy comprehensive evaluation and back propagation(BP)neural network.The comprehensive evaluation model considers key dimensions such as topological quality,geometric features,and coordinate errors.It enhances evaluation accuracy and objectivity by optimizing subjective and objective weights through a combination of fuzzy comprehensive evaluation and BP neural network.The experimental results indicate that the proposed method provides fidelity evaluation results for watermarked vector maps that more accurately reflect the actual data,offering a comprehensive and objective framework for evaluating digital watermarking algorithms,which is more scientifically robust and rational than conventional error analysis approaches.Furthermore,this method is significant in determining the optimal watermark embedding strength,and can guide the adjustment of watermark algorithm parameters according to different application scenarios and data characteristics,providing a basis for balancing the comprehensive performance of watermarking algorithms.