基于DCT系数与压缩感知的图像哈希算法

Robust Image Hashing Based on DCT Coefficient and Compressive Sensing

提出一种基于DCT系数和压缩感知的图像哈希算法。先将输入图像规格化,随后进行DCT,取得其第一列和第一行系数;然后对两组系数进行压缩感知得到测量向量,计算其均值得到哈希值。ROC曲线对比验证表明,该算法分类性能更优。

In this paper,we propose a perceptual image hash function with compressive sensing and DCT coefficient.Concretely,the input image is mapped to a normalized image by prepossessing.On the basis of prepossessing,we extracts DCT coefficient to construct feature matrices.Next,compressive sensing basis is applied to each vector and finally the mean of each vector are used to form hash vector.Experiments show that the proposed algorithm reaches good discrimination.Receiver operating characteristics（ROC）curve comparisons indicate that the proposed hashing algorithm is better than some notable image hashing.