基于人类视觉系统的水印容量分析

WATERMARKING CAPACITY ANALYSIS BASED ON HUMAN VISION SYSTEM

数字水印容量是指在满足不可感知性和鲁棒性的前提下所能隐藏信息的数量。旨在人类视觉系统的特性的基础上,分析水印的容量。先对图像进行离散余弦变换,然后由Watson的视觉掩蔽模型计算每个系数的临界可见误差,即最大可修改幅度,在这里对这个最大幅度值进行了调整,最后利用多比特估计方法得到了水印容量的计算公式。随后对标准图像进行了实验测试,得出了调整因子与容量、峰值信噪比和平均嵌入量的关系。只要确定了峰值信噪比,调整因子和水印容量就确定了。

Digital watermarking capacity refers to the amount of bits for hidden image information which satisfying the condition of imperceptibility and robustness. The purpose of this paper is to analyze the capacity of digital image watermarking based on the character of human vision system （HVS）. The discrete cosine transform （DCT） was done on image first, then Watson visual masking model was used to calculate just noticeable difference （JND） , i.e. maximum revisable range, of each parameter-the maximum range value is adjusted in this paper. Finally the calculation formula of watermarking capacity was derived on the use of multi-bits estimation algorithm. After that the experimental tests were carried on with standard images, the relation graphs were obtained between adjustment parameter and capacity, and between peak signal noise ratio （PSNR） and average embedded number. When PSNR is determined, the adjustment parameter and watermarking capacity will be determined as well.