数字图像取证技术研究进展综述

THE RESEARCH PROGRESS OF DIGITAL IMAGE FORENSICS: A REVIEW

本文着眼于数字图像取证技术的研究成果,对数字图像取证技术的研究进展进行了系统的论述.从光学一致性、传感器规律一致性、图像篡改时留下的痕迹或成像时留下的痕迹以及统计特征一致性四个方面对基于内在一致性的传统数字图像取证技术进行了概述;从简单的模型迁移、对模型输入层的修改以及模型架构的修改三个方面对深度学习的图像取证技术进行了阐述.最后,本文对现在数字图像取证技术存在的挑战进行了概括,并展望了取证技术的发展方向.在各种媒体环境中充斥着效果逼真的伪造图像的环境中,对图像取证技术的了解可以让人们更加理性地认识“眼见为实”的观念,提高人们的媒体安全意识.

This paper focuses on the research of digital image forensics,of which the research progress is introduced briefly.The traditional digital image forensics is based on internal consistency,and its research is summarized in four aspects:optical consistency,sensor regularity consistency,image tampering or imaging traces,and statistical characteristics consistency.The advanced image forensics is based on deep learning,and here it is introduced in three aspects:the transfer of the simple model,the modification of the model input layer,and the modification of the model architecture.Finally,the challenges of digital image forensics are summarized and its development directions are discussed.In the media environment with various fake images,the understanding of image forensics technology can make people more rational understanding of the concept of"seeing is real"and improve the media security awareness of people.