多幅混合数字图像提取的迭代学习方法

Iterative Learning Method for Extraction of Digital Images Under Multi-blending

提出以迭代学习辨识算法提取隐藏图像的新方法.将图像信息嵌入到混沌系统的参数中,并运用图像多重混合嵌入技术完成图像隐藏.这种图像隐藏过程等效于含未知时变参数的动态系统.运用D型迭代学习辨识方法在有限时间区间上对时变参数的完全估计能力,实现隐藏图像信息的完全提取.数值结果表明在各种攻击下该提取方法的有效性.

This paper presents a novelrecovery scheme of the hidden image, from iterativeleaxning identification perspective, where digital image information hiding scheme is on the basis of chaotic mapping and multi-blending. The digital image information is added to a parameter of the chaotic system undertaken, an image can be hidden in one carrier image through multi-blending with multiple images. Iterative/earning identification method can be used to give complete estimation of time-varying parameters over a finite time interval, and a D-type learning algorithm is proposed for the recovering purpose. Numerical results demonstrate the effectiveness of the proposed learning algorithm, by which the digital image can be recovered completely, and it is able to resist various types of attacks.