Watermarking Algorithm Analysis And Research Based On Wavelet Transform And Singular Value Decomposition

Nowadays, only with traditional encryption technology, it’s hard to protect theDigital products. As a supplement to the encryption technology, information hidingtechnology is paid more and more attention to, and digital watermark is the hot spot ofthe application.The digital watermark is divided into two types: the spatial domain and transformdomain watermark, and for the former one, the watermark information is directly addedon to the unimportant bit of the grey value, so robustness is quite limited. Based on thethought of transform domain, this paper will use wavelet transform for imagetransformation method and design a digital watermarking system. Discrete wavelettransform has very good time-frequency localization, and concentrates main energy ofimages on the low frequency components, which is adapted to the human perceptualsystem and compatible with JPEG2000. However, to overcome the shortcomings thatthe wavelet transformation does poor at resisting the geometric attack, the paper willregard the singular values of the carrier and the watermark as the main body of thewatermarking system, because the singular values of the image were constant, even ifsuffered from of geometry distortion. At the same time, in order to guarantee thesecurity of watermark, this paper will do Arnold scrambling and chaotic encryption towatermark image to eliminate image pixel grayscale value of space correlation. Tofurther improve the ability to resist geometric attack, for rotation attack, this paper willdetermine the image rotation angle with the image normalization; for shear attack, thelow frequency components will be divided into several pieces of44, before singularvalues decomposition.Through the analysis of the experiment, the watermark information is concealedwell, and the watermark system has good robustness, and can resist the image processintentionally or not.