Algorithm Research On Color Image Watermarking For Copyright Protection

As an important branch of research in the field of information security, Digital watermarking technology was an effective digital copyright protection method. Digital watermarking was the technology that does not affect the use of digital carrier, and it was not perceive the human visual system easily, it contained specific meaning identification information (digital watermark) using data hiding algorithm embedded into a digital carrier, by these hidden information in the carrier could achieve the goal that confirming the contents of the original creator, or determine whether the carrier had been tampered. This paper was based on research which conducted in the color image digital watermarking technology in its copyright protection method. It mainly included the following aspects:(1) An algorithm was presented in this paper which was based on multi-channel biorthogonal wavelet transform for embedding a color digital image into an original color image. The algorithm combined the well characteristics of decomposition and reconstruction of biorthogonal wavelet. After the Arnold transform to the watermark image, the carrier and the watermark image were respectively into two and one level biorthogonal wavelet transform. The position and intensity of the watermark embedding was decided according to the weighting coefficients corresponding to different frequency bands and characteristics of the human visual system, the watermark with corresponding strength was embedded into the carrier image. The embedded watermark image got through the inverse biorthogonal wavelet transform. According to the experiment, the algorithm had invisibility and good robustness for common image processing operations attack.(2) An algorithm was presented which was based on NSCT and PCA for embedding a color digital image into an original color image. Algorithm uses a good shift invariance of NSCT and advantages of compressing data image information of PCA. The procedures were to put RGB watermark and carrier image converting into YUV image, then as to Y channel, the carrier and the watermark image were respectively into two and one level non-sampling contourlet transform, and then put matrix into matrix blocks, the low and high frequency coefficients of the blocks with rich texture information, respectively using PCA to extracted principal components, and then different intensities were used for embedding. Experimental results showed that the algorithm had high robustness and good invisibility.