Abstract:
In the past decade, we have seen a remarkable growth in ability to capture and distribute digital videos. But, taking into consideration the
availability of too many powerful and user-friendly video editors, the question is how to verify the authenticity of these digital videos. The
problem becomes crucial, when it comes to videos used in courts as evidences or journalism. Based on our deep experience in the field of
digital image forensics and cooperating with law enforcement authorities, our aim is to develop several highly needed methods for identifying
tampering in digital videos as well as ballistic analysis of video signals. Specifically, we will develop methods for video source verification
using adopted sensor noise patterns, provide an extensive quantitative experiment analyzing their robustness and detect video tampering
using sensor noise and non-consistencies in traces of interpolation. Moreover, we propose a method detecting re-captured videos.