Detecting Frame Deletion in H.264 Video

Abstract

Frame deletion is one of the common video tampering operations. The existing schemes in detecting frame deletion all focus on MPEG. This paper proposes a novel method to detect frame deletion in H.264. We introduce the sequence of average residual of P-frames (SARP) and use its time- and frequency- domain features to classify the tampered videos and original videos. Specifically, in the time domain, we analyze the periodicity of the SARP of videos with frame deleted and define a position vector to describe this feature. In the frequency domain, we demonstrate that the periodicity of SARP results in spikes (frequency-domain feature) at certain positions in the DTFT(Discrete Time Fourier Transform) spectrum. The time- and frequency- domain features of tampered videos are different from that of original videos and thus can be used to separate these videos apart. Experimental results show that the proposed method is very effective with the detection rate as high as 92%.