Adaptive Zero Watermarking Algorithm Based On NMF And Boost Normed Singular Value

An adaptive zero watermarking algorithm based on block non-negative matrix decomposition(Non-negative Matrix Factorization,NMF)is proposed to resolve the problem of high false alarm rate and weak robustness,which was caused by Singular Value Decomposition(SVD).First,using the Arnold transform and Logistic mapping to apply to the watermarking image,and the encrypted watermarking image is obtained.Then,the original gray image is pre-processed by Discrete Wavelet Transform to transform into two levels of DWT,and then the low frequency sub-bands are divided into non-overlapping blocks,and each sub-block is decomposed by NMF with rank r which the aim at obtaining the decomposed sub-band.The characteristic matrix is decomposed by Boost Normed Singular Value Decomposition(BN-SVD).The feature vector is constructed by the relationship between the maximum singular value of each block matrix and the mean value of the global maximum singular value.Zero watermark is created by feature vector and encrypted watermark image are XOR operation.By using the global optimization of the Beetle Antennae Search(BAS)algorithm to determine the parameters in the boost normed value decomposition to solve the diagonal distortion problem.By using the normalization coefficient can assess the watermark performance criterion.The data show that the algorithm is resistant to JPEG compression,noise,filtering,rotation,shearing and hybrid attack.The NC value of original image and extracted watermarking image can reach 0.99.The scheme is not only solving the false alarm problem efficiently and improved the robustness substantially and it also can resist all kinds of attacks effectively.