基于Bandelet与自适应矩阵范数的全频率无损水印方法

数字水印技术是一种新型的数字图像版权认证和保护技术.传统的数字水印一般通过修改图像数据达到信息隐藏的目的,不适合用于保护不允许修改的图像数据,无损数字水印为这类数据的保护提供了一种有效的方法.提出了一种无需对图像数据做修改的全频率无损数字水印实现方法.首先对图像实施小波变换,变换后的中高频部分采用Bandelet跟踪图像几何流,将纹理和边缘作为图像的特征从而构造出图像高频无损水印参数.对于小波变换的低频部分,通过选取最优的矩阵范数,得到低频系数水印参数生成方案,达到提取图像统计和边缘特征,对图像实现全面保护的目的.实验证明该算法具有良好的抗攻击能力,能够广泛用于对不可修改数据的保护.

A Complete Frequency Lossless Watermarking Method via Bandelet and Adaptive Matrix Norm

The watermark is the new technology which can be used to apply to authenticate the image copyright and offer the safety. The traditional watermark protects the digital image through modifying the image data to hide the information which is used to authenticate the image copyright. It does not suit protecting some images which are not permitted to be destroyed. The lossless watermarking is an effective method to protect these image data. A complete frequency lossless watermarking method with which the image data need not be changed is proposed in this paper. Wavelet transform is first performed for the original image, and then for the high and middle frequency part of the image transformed, the geometric flow is traced by using Bandelet, and the texture and edge considered for feature parameters of the image are used to construct high frequency lossless watermarking. For the low frequency part of wavelet coefficients, by selecting optimal matrix norm a new watermarking scheme is then proposed. Through drawing the statistical character and the edge of the image, the method is able to protect the image from allaround attack. The experimental tests show that the proposed approach has the capability to resist strong attacking, and it can be widely used in protecting data which are not permitted to be modified.