一种用于遥感影像完整性认证的感知哈希算法

Perceptual hash algorithm for integrity authentication of remote sensing image

为了实现遥感影像的完整性认证,提出了一种基于边缘特征的遥感影像感知哈希算法.首先,对影像进行格网划分,并对格网单元进行自适应的预处理.其次,采用Canny算子提取格网单元的边缘特征,进一步生成边缘特征矩阵.然后,对边缘特征矩阵进行奇异值分解,选择较大的奇异值作为格网单元的特征,并采用Hash函数进行归一化.最后,串联所有格网单元的归一化特征,得到影像的感知哈希散列.在影像接收端,计算待认证影像的感知哈希散列,并与收到的感知哈希散列进行匹配,实现遥感影像的完整性认证.实验结果表明,该算法能够有效检测遥感影像的局部细节篡改,对于保持影像内容不变的操作具备较好的鲁棒性,且运行效率较高,为遥感影像的内容完整性认证提供了一种高效途径.

In order to authenticate the content integrity of a remote sensing image, a perceptual has-hing algorithm based on edge feature is proposed.First, the remote sensing image is partitioned into grids, and all the grid cells are pretreated adaptively.Secondly, the edge features of the grids are ex-tracted based on the Canny operator to construct the edge feature matrix.After singular value decom-position （ SVD） on the matrix, the larger singular values are selected as the features of grid cells, which are further normalized by the Hash algorithm.Finally, all the normalized features of the grid cells are connected in series to construct the perceptual Hash value.At the receiver, the authentica-tion process is implemented via the comparison between the reconstructed perceptual Hash value and the original one.The experimental results show that the proposed algorithm can effectively detect the local detailed tampering of the remote sensing image with a high running efficiency, and has good robustness to content-preserved operations.Thus, the method realizes an efficient content integrity authentication for remote sensing images.