一种改进的DWT域图像篡改检测算法

改进一种DWT域图像篡改检测算法。该算法将置乱的有意义的二值水印图像利用量化的方法隐藏在载体图像Haar小波变换系数中。图像认证时,对提取的水印和原水印图像的差值图像进行反混沌置乱,再进行形态学处理,从中可以看出认证图像的篡改区域。此算法与Kunder等人提出的基于Haar小波的半易损水印算法相比,能够有效地区分JPEG压缩和恶意篡改,不需要设置阈值来区分JPEG压缩和恶意篡改,可以从水印差值图像直接看出恶意篡改的区域。

An improved tamper detection algorithm Based on DWT

This paper improves a DWT Image tamper detection algorithm.The algorithm uses a meaningful binary image as watermark and applies quantitative methods to hide the watermark image in coefficient vectors of Haar wavelet transform. In order to detect collage attack, regional carrier is confused by chaos before hiding. When image authentication, marginal image between extracted watermark and the original watermark is applied re-chaos transform and processed morphological processing which is to eliminate scattered points brought by JPEG compression or general image processing. Consequently we can detect tampered region of authentication image from processed marginal image. Contrasted this algorithm with Kunder's algorithm which is a semi-fragile watermarking algorithm based on Haar wavelet transform, this algorithm is able to effectively distinguish between JPEG compression and malicious tampering. Malicious tampered region can be seen directly in the processed marginal image.