Research On Digital Watermarking Algorithm And Evaluation Method Based On Wavelet Transform

In recent years, the security problem of the information system is graduallyincreased, information security is very important to the medical, aviation, nationalpolitics, military and so on.With the development of digital watermarkingtechnology, how to hide the information, and to get better extraction effect in avariety of channel attack, become the direction of experts in various fields.There are three main kinds of digital watermark embedding, watermark infrequency domain,spatial domain and compressed domain. the watermark infrequency domain including Fourier transform, discrete cosine transform, andwavelet transform. Because of time / frequency characteristics and multi-resolutioncharacteristics, the wavelet transform has been widely used in the field of digitalwatermarking.This paper proposes a digital watermarking method based on wavelet transformwhich use the one-dimensional stochastic chaotic system to determine the watermarkembedding position, and use two-dimensional stochastic chaotic system toscrambling watermarking. Then, the horizontal high frequency wavelet transformare replaced by the corresponding vertical high-frequency component. By this way,which can ensure the invisibility of the watermark and improves the anti attackcapability. Through the simulation experiment, we measured the algorithm can resistmost of the attacks, especially has a good effect on the shear attack and luminancetransformation attack. Due to color image accounted for a large proportion in thedaily communication, we then apply this algorithm to color image.Because of the limitation of traditional watermark imperceptibility evaluationmethod, this paper improves SSIM algorithm which were extracted from thelow-frequency and high-frequency components of different resolution by usingmulti-level wavelet transform, then each components are weighted by CSF curve. Inorder to improve the measurement accuracy, we carry on DCT transform in lowfrequency component, then measured the high frequency component value of SSIM,and multiply it to the measured value to get the final detection value. For colorimages, taking into account the color masking effect of human visual, we calculationR, G and B of each channel then weighted them. Through the experiment of gray andcolor images, we found that this algorithm has a higher fitting degree, that is to saythis algorithm is more similar with human subjective observations. Finally thewatermark imperceptibility evaluation method is used to evaluate the watermarkembedding algorithm and achieved good results.