Research On Image Authentication Based On Hash Function Active Forensics And CFA Interpolation Passive Forensics

With the rapid development of computer network technology and information technology,the acquisition of digital images has become more and more convenient and the transmission channels have become increasingly diversified.At the same time,various image editing and processing software have also rapidly developed,making it easy for people to modify,modify,and even tamper with a digital image.These pictures are gathered in various media such as the Internet,newspapers,magazines,etc.,and bring many unpleasant factors to personal privacy,economic development,and social stability.They have extremely bad influence on the country's politics,economy,and culture,and even initiate international disputes.In this paper,after a thorough study of the current state of all kinds of digital image active and passive forensics technology and its research status at home and abroad,we first study the representative fragile watermark active forensics technology,and propose a Hash function based on Henon mapping.The gray-scale image active forensics method elaborated the fragile watermark generation,embedding,and the whole process of image tampering and positioning,and was simulated on MATLAB.The experimental data show that the algorithm has good visual invisibility,has high sensitivity to the tampering of the watermark image,and can be accurately positioned,and can be used to identify the correctness and completeness of the gray-scale image.After analyzing the pros and cons of active forensics technology,a more in-depth study of CFA interpolation passive forensics technology with certain innovative value was conducted.Focusing on the principle of digital camera imaging and several commonly used interpolation algorithms,two passive forensics algorithms based on CFA interpolation are proposed for color images.The first is to first construct the interpolated pixels of the test image and obtain the prediction error of the pixels.The weighted variance function is used to establish the Bayesian probability model,and the likelihood ratio is used to define a method of outputting the forgery mapping of the tampered region.The second method is to estimate the "original interpolation mode image" and combine the Canny edge detection and the expansion filter to position the splicing tampered image.Simulated by MATLAB respectively,the experimental data shows that both methods have relatively clear positioning of the tampering area of the color image: The former is accurate for copy-paste tempering of common images and different images,especially for operations such as rotation and scaling.The tampered region of the post-erase CFA interpolation feature is particularly sensitive,enabling fine-grained pixel-level positioning,but the visual effect is less obvious;while the latter shows only the edge information of the stitched region,the algorithm is robust and robust.The degree of JPEG compression,filtering,and noise processing can accurately locate the tampered area.However,the edge information cannot directly tell what part is the tampered area,and needs to be subjectively distinguished by the human eye.Finally,it summarizes the two types of forensic techniques studied in this paper,points out the inadequacies and improvements,and looks into the direction of future research.