A DCT-SVD Domain Watermarking for Color Digital Image Based on Compressed Sensing Theory and Chaos Theory

Abstract

In this paper, it is presented that a new color image watermarking algorithm based on compressed sensing theory and chaos theory in discrete cosine transform (DCT) and singular value decomposition (SVD) domain. Initially, a color watermarking image is preprocessed by four steps i.e. Being divided into red, green and blue channels, being sparse by wavelet transform, scrambling by 1-D chaos sequence and measurement through Gaussian random matrix. Here, the chaos sequence is produced by 1-D Logistic mapping. Then the red, green and blue channel images of a color host image are divided into 8×8 blocks and then transformed by DCT. Next the first DCT coefficient of each block is selected to compose a new matrix, and the new matrix is decomposed by SVD. The singular values are the regions of watermarking embedment. Thirdly, three sets of measured coefficients as new watermarking are embedded into the corresponding singular values. Finally, the Iterative Hard Threshold (IHT) algorithm is used to reconstruct the watermarking image. The method proposed can not only enlarge the information capacity of watermarking embedment but also strengthen the security of watermarking in the condition of the good imperceptibility. The experiments show that the scheme is feasible and has good anti-attack ability for JPEG compression, Gaussian white noise, rotating and median filtering.