An HEVC-based known-plaintext attack for video selective encryption

In this paper, we propose an HEVC-based known-plaintext attack (HKPA) to reveal and evaluate these weaknesses. Unlike typical KPAs, our approach assumes the attacker has access to only a single plaintext-ciphertext video pair, which makes cryptanalysis considerably more challenging. We first define the slice synchronization identifier and then design a decoder for the synchronized slice to generate the estimated key sequence (EKS). Once the attacker acquires the EKS, it becomes possible to decrypt other videos encrypted with the same initial key. Our experimental findings demonstrate that a significant portion of the video content in decrypted videos is recognizable, with the EKS achieving an average accuracy exceeding 86%. These results underscore the vulnerability of current SE schemes to our proposed HKPA.