一种基于特征提取与追踪的视频复制粘贴篡改检测方法
A copy-paste video tamper detection method based on features extracting and tracking

提出一种基于特征提取与追踪的视频篡改检测方法. 首先对视频序列按照一定帧间隔进行特征提取，寻找出首个篡改帧及其篡改区域；其次采用追踪算法对该篡改区域进行学习，并定位该帧之后的每个视频帧的可疑区域；最后对可疑区域再使用特征提取算法进行相似点匹配，确定出视频篡改帧的位置及其篡改区域. 结果证明了方法的有效性和鲁棒性，以及在执行时间上相对于已有算法的优越性.
In this paper，a novel video tamper detection method is proposed. In the method，the SIFT algorithm is firstly used to extract features of video sequence according to inspection interval parameter and seek out the first tampering frame and the tampered area. Then，the compression tracking algorithm is used to study the tampered area and locate the suspicious area of each video frame. Finally，the SIFT algorithm is used to calculate the similarity of suspicious area obtained and the position of tampered frames of the whole video and their specific tampered areas are confirmed. The experimental result show that this method are more effective，robust and quick than that of before