Video Authentication Using Relative Correlation Information

This paper presents an intelligent video authentication algorithm based on relative correlation information. It computes the local relative correlation information between all the adjacent frames of a video and projects them into a nonlinear SVM hyperplane to determine if the video is tampered or not. Performance of theproposed algorithm is not affected by acceptable video processing operations such as compression and scaling and effectively classifies the tampered videos. On a database of 640 videos, the proposed algorithm yields a classification accuracy of 99.0%.