Research On Digital Watermarking Technology Based On Image Structural Characteristic

With the rapid development of network technology,multimedia data increases with the explosive exponential growth.The issue of protecting the security of multimedia data has become a focus of attention in the field of information security.As an effective way to protect the security of multimedia data,digital watermarking has been used in many fields such as multimedia copyright protection,data authentication,fingerprint recognition and multimedia indexing.As a classical watermarking algorithm,quantization-based watermarking schemes can embed the watermark bits by quantizing the host signal samples with a set of quantizers.It has received widespread attention and provides an efficient watermarking algorithm with high data capacity.In the quantization-based watermarking framework,the perception characteristics of human visual system have a great influence on the visual quality of watermarked image.In recent years,the quantization-based watermarking algorithm via visual perception model can achieve a better tradeoff between fidelity and robustness.In the processing of perceive the scene,the human brain can adaptively extract the structural regularity from the scene.However,the existing JND(just noticeable difference)models which are used in quantization-based watermarking framework do not consider the impact of structural characteristics of image,so it can not accurately estimate the actual visual perception redundancy.In this paper,the impact of structural characteristic on visual masking effect is considered.An effective visual JND model is proposed,which is applied into the quantization-based watermarking framework.The contributions have been made in the following:1? A new method for block classification is used to classify DCT blocks,which can effectively distinguish the regular texture regions from the region with high texture complexity.A new method is used to calculate the edge density, which measure the texture complexity of a block along different directions.2? A new JND model is proposed,which describes the perception characteristics more effectively.The proposed block classification method is used to estimate the contrast masking effect.Consequently,the contrast sensitivity function and the luminance adaptation effect are also considered to calculate the JND model.3? Based on the proposed JND model,an improved STDM(spread transform dither modulation)watermarking algorithm is proposed.Experiments show that the proposed scheme achieves better tradeoff between fidelity and robustness and outperforms other STDM schemes.