Confidentiality of a selectively encrypted H.264 coded video bit-stream

Transparent encryption with scalable video communication: Lower-latency, CABAC-based schemes

“…Hence, it might be claimed that it is easy to guess the values of the changed signs and, thus, make the video watchable. In addressing this issue, consider Table 2 of [25]. In [25] it is demonstrated that there are literally millions of MVDs and TCs in some of the video sequences listed in Table 9 of the current paper, which video sequences also appear in Table 2 of [25].…”

“…In addressing this issue, consider Table 2 of [25]. In [25] it is demonstrated that there are literally millions of MVDs and TCs in some of the video sequences listed in Table 9 of the current paper, which video sequences also appear in Table 2 of [25]. The probability of guessing the signs [25] can be found from the standard formula for a combination:…”

“…Because the number of signs that would need to be guessed successfully is so large, the probability of guessing all the signs is very low. In [25] it is demonstrated that if the Bus sequence is taken as an example, the number of ways of guessing MVDs signs with complete accuracy is 2^119904. In [25] it is also shown that even if 80% of the MVD signs were guessed successfully the Bus video would still not be in a watchable condition.…”

“…In [25] it is demonstrated that if the Bus sequence is taken as an example, the number of ways of guessing MVDs signs with complete accuracy is 2^119904. In [25] it is also shown that even if 80% of the MVD signs were guessed successfully the Bus video would still not be in a watchable condition. Therefore, the proposed sign changing method is sufficiently secure without applying a standard encryption method such as AES.…”

“…In both these methods, the decoder bitstream remains format compliant. Other work by us has confirmed the resistance to perceptual attacks [25] and examined to what extent the core SE technique can tolerate errors in a wireless channel [26].…”