基于图论的视觉显著模型的图像哈希算法

Image Hashing Algorithm Based on Graph Theory and Visual Saliency Model

近年来,图像哈希算法在多媒体安全领域引起了广泛的关注,已成功应用于社会事件检测、复制检测、数字水印等领域,但许多算法在鲁棒性和唯一性之间没有达到很好的平衡,针对这个问题,本文提出了一种基于图论的视觉显著模型的图像哈希算法。该算法首先通过GBVS模型构造预处理后图像的视觉显著图,然后通过Logistic混沌映射生成位置映射数组,最后对数组数据进行哈希加密并得到最终的哈希值。利用公开的数据集去验证该算法的性能,结果表明该算法可以抵抗常见的数据操作,其分类性能也优于某些文献算法。

Image hashing has attracted much attention of the community of multimedia security in the past years.It has been successfully applied to social event detection,copy detection,digital watermarking and so on.But many algorithms do not achieve a good balance between robustness and discrimination.To address this problem,this paper proposes an image hashing algorithm based on a visual saliency model of graph theory.The algorithm constructs the visual saliency graph of the preprocessed image by the GBVS(Graph Based Visual Saliency,GBVS)model,and then generates the position mapping array by logistic chaos mapping,hash encryption of array data,and gets the final hash values.The performance of the algorithm is verified using publicly available datasets,and the experiments demonstrate that the algorithm can resist common data manipulations and outperforms some algorithms in literature for classification.