A robust VVC video encryption scheme based on the dynamical chaotification model

Abstract

With the considerable surge of video assets, the demand for video content security has undergone substantial growth. Since the advent of H.264/AVC and H.265/HEVC, selective encryption algorithms (SEAs) have become a focus. However, there is little research based on the H.266/VVC. To fill the gap, a VVC encryption scheme is proposed in this paper. Considering that most existing SEAs are easily cracked under cryptographic attacks, a dynamical chaotification model (DCM) is designed to resist chosen-plaintext attack (CPA). Additionally, given that a majority of existing SEAs are incapable of decrypting correctly in the case of packet loss, each slice is encrypted independently to enhance robustness. Firstly, the three-dimensional (3D) DCM is designed based on the exponential feedback and the boundary function. Secondly, the dynamical parameter in the 3D-DCM is updated by the encoding parameters in the VVC encoder. Then the initial values and other model parameters of the 3D-DCM are scrambled by the hash values of the slice headers (SHs). Finally, the texture, motion, and structure information of VVC videos are encrypted using the iterative output of the 3D-DCM. In particular, three exclusive elements in H.266/VVC are selected in the RSES, including matrix-based intra prediction (MIP), cross-component linear model prediction (CCLM), and multiple transform selection (MTS). Experimental results demonstrate the outstanding performance of the proposed RSES in robustness and resistance to CPA, while gratifying bit rate and real-time requirements. Therefore, it is a promising encryption scheme in cryptographic applications.

Keywords

• Chaotification model
• H.266/VVC
• Video encryption
• Robustness