Research On Digital Image And Video Copy-Move Tampering Forensics

With the rapid development of Internet and multimedia technology,editing and modifying the image or video are becoming increasingly easier.As the most intuitive carrier of information dissemination,digital image and video are applied to various fields such as judicial justice,medical claims,news photography,scientific research and national security.Digital images or videos that have been malicious tampered will undoubtedly have a serious negative impact on the society and the country.Therefore,the research on digital image and video tampering forensics has become a research hotspot in the field of information security.Due to the diversity of tampering intentions and editing software,people can easily tamper with the images and videos in many different ways.The copy-move forgery is one of the most simple and common way of manipulations.This thesis thoroughly researches the copy-move tampering forensics method of digital image and video,and the thesis's contribution mainly concentrates on the following three aspects:1.To detect the copy-move forgery in a digital image,a detection algorithm based on local intensity order pattern is proposed in this thesis.Firstly,the local intensity order feature descriptors are exacted from the uninspected image.Then,the angular cosine of feature descriptors is calculated to match the features,and the stable matching points are found according to the distance ratio threshold of the nearest neighbor point to the second nearest neighbor.Finally,the space distance of the matching points is calculated to remove the false matching points.Extensive experimental results confirmed that the proposed method is not only able to effectively identify and locate the altered area,but also has high accuracy and is robust to many operations of post-processing,such as scaling,rotation,brightness change,Gaussian blur,additive white Gaussian noise and JPEG compression.2.To detect the continuous multiple copy-move forgery in videos' single frame,a detection algorithm using quantized Discrete Cosine Transform coefficients is proposed in this thesis.Firstly,the uninspected video is converted into sequential frames.Then,the quantized Discrete Cosine Transform coefficients are taken as the feature vector of a frame image,and the similarity between adjacent frames is measured by calculating the Bhattacharyya coefficient,then the suspected frames with abnormal similarity will be found by the set threshold.Finally,according to the continuity of suspected frames and the continuous frames to determine whether the video is tampered,and locate the tampered position.Experiment results showed that the proposed algorithm can detect the video with different scenarios,and has advantages of high speed detection,free from the influence of compression,high accuracy and low omission ratio.3.To detect the up-conversion tampering of video frame-rate,a detection algorithm based on the periodic properties of optical flow is proposed in this thesis.Firstly,the video is converted into sequential frames.Then,the Horn-Schunck optical flow method is used to calculate the optical flow vector of pixels in the horizontal direction and vertical direction between adjacent frames,and then the optical flow change rate of the adjacent frames will be calculated.Finally,the fast Fourier transform is used for the spectrum analysis of the optical flow change rate,and by judging whether the ratio of the max amplitude in the spectrum to the mean amplitude is higher than a certain threshold,whether the video is altered by frame-rate up-conversion will be confirmed.Experiment results showed that the proposed algorithm not only can accurately identify the video tampered by frame-rate up-conversion,but also improve the robust performance of video compression well.