基于NMF和Hu不变矩的遥感影像零水印算法

A Zero Watermarking Algorithm for Remote Sensing Images Based on NMF and Hu Moments

为满足遥感影像数据在遭受大幅裁剪(强裁剪)后仍能有效保护版权的需求,该文提出一种基于非负矩阵分解(Nonnegative Matrix Factorization,NMF)和Hu不变矩的零水印算法。首先将原始影像数据分解为系数矩阵和基矩阵,然后计算整体Hu矩不变量均值和分块不变量均值,将两者得到的特征值二值化后与版权信息进行异或操作以生成零水印。其中,系数矩阵及零水印需存入第三方机构,为版权验证奠定基础。在检测版权信息时,利用NMF可复原特性,从待检测数据中反算出原始数据,并提取零水印以进行版权信息比对。实验结果表明,该算法具有版权保护的唯一性和鲁棒性,尤其在抗强裁剪方面表现突出,显示出在遥感影像数据版权保护领域的应用价值。

Aiming at the demand of copyright protection for remote sensing image data with strong cropping resistance,this paper proposes a zero watermarking algorithm based on the nonnegative matrix factorization(NMF)and Hu moments.The algorithm first decomposes the original image into coefficient matrix and basis matrix,then calculates the overall Hu moment invariant mean value and chunk invariant mean value,and binarizes these feature information and performs the dissimilarity operation with the copyright information to generate the zero watermark.The coefficient matrix as well as the zero watermark need to be deposited in a third-party organization.When detecting copyright information,the original data is back calculated from the data to be detected by utilizing the recoverable characteristics of NMF,and the zero watermark is extracted for copyright information comparison.The experimental results show that the proposed algorithm has uniqueness in copyright protecti on and robustness,especially in the resistance to strong cropping,showing its application value in the field of copyright protection of remote sensing image data.