基于视觉模型与菱形编码的图像隐写算法

Image steganographic algorithm based on human vision system and diamond encoding

为提高图像空域隐写算法的视觉不可感知性,通过结合视觉注意力机制和人眼对图像的灰度及边缘等敏感特性,提出一种基于视觉模型的感兴趣区域菱形编码隐写算法.在底层人眼感知判定的基础之上,根据视觉注意模型划分嵌入等级,并动态调整秘密信息的嵌入量.实验结果表明,与基于边缘自适应的像素对匹配算法相比,本文提出的算法不仅有较高的视觉不可感知性,而且能抵抗常用的空域隐写分析.

To improve the invisibility of image spatial steganography,we present a ROI-APPM algorithm ofinterest region by combining vision mechanism and the eye sensitivity to gray and edge.The embedded secretinformation is dynamically adjusted through the visual attention model and low level perceptual judgment of thehuman eyes.Experimental results show that,compared with EAPPM,the proposed algorithm not only achievesa higher visual sensitivity,but can also resist the traditional spatial domain steganographical technique.