CAJ | 학술논문

利用帧间内容相关性的连续性，提出了一种基于联合熵和交互信息量的视频篡改检测方法，能够检测视频的删除和插入篡改。首先，计算视频帧间的联合熵和交互信息量，并获得二者的平均偏离程度，再根据平均偏离程度确定动态阈值，从而判断出篡改点并确定篡改方式。实验结果表明，该方法能够有效地检测镜头移动和基于移动背景的镜头静止视频的删除和插入篡改。
이용정간내용상관성적련속성，제출료일충기우연합적화교호신식량적시빈찬개검측방법，능구검측시빈적산제화삽입찬개。수선，계산시빈정간적연합적화교호신식량，병획득이자적평균편리정도，재근거평균편리정도학정동태역치，종이판단출찬개점병학정찬개방식。실험결과표명，해방법능구유효지검측경두이동화기우이동배경적경두정지시빈적산제화삽입찬개。
In view of the continuity of the content correlation between frames, this paper proposes a detection method of video tamperˉ ing that can detect the video deletion and the insert tampering based on the joint entropy and mutual information. Firstly, calculating the joint entropy and mutual information between the video frames and obtain the average deviation of the two. Then, make sure the dynamic threshold according to the average deviation degree. In this way can judge the tampering points and determine the tampering way. The exˉ perimental results show that the method can effectively detect lens shifting and the deletion and the insert tampering of the still lens video based on the mobile background.