Encryption Algorithm Of Spatial Scalable Video Based On SHVC

HEVC has significant compression performance and is widely promoted in high resolution video applications.As its scalable and extended version,the SHVC standard has attracted much attention in the fields of streaming media services,video conferencing,and public video surveillance because it can dynamically adapt to different network environments and user terminal devices.The existing encryption algorithm based on scalable video has few achievements.Therefore,this paper focuses on the SHVC compression coding standard to carry out the corresponding encryption algorithm research work and propose scalable video encryption algorithms for HD resource copyright protection and personal privacy information management.(1)A scalable video encryption algorithm based on interlayer processing is proposed to solve the problem of insufficient visual security in the protection of SHVC enhancement layer(EL)video.On the basis of the intra-layer texture or motion information encryption,the encryption algorithm is combined with the inter-layer prediction process,which greatly reduces the video quality of the EL and preserves the compatibility.The SHVC contains three different inter-layer processing modules.In order to compare and analyze the encryption performance,respectively,three encryption algorithms are proposed.The experimental results show that the algorithms have high security,stability and anti-attack performance.The average reduced PSNR is greater than20 dB.SE-SHVC-C and SE-SHVC-M introduced compression load are not more than 1%.(2)Based on the compressed domain,the SHVC EL video encryption algorithm is proposed,which has compression ratio invariance and format compatibility.Based on the analysis of SHVC code stream structure,the algorithm demultiplexes the EL data to be encrypted,and extracts the suitable slice information for encryption.The algorithm encrypts the coding unit(CU)partitioning mode,the candidate list and reference picture index information in inter prediction,the absolute values and symbols of motion vector differences(MVD)and non-zero quantized transform coefficients(QTC)after partial entropy decoding.Finally,the conflict avoidance is adopted.Experimental results confirm that the algorithm can effectively scramble the EL video,has strong stability,can resist exhaustive and plaintext attacks,does not introduce compression load,the average encryption time per frame is about 0.17 s..(3)To improve the security of privacy in public video surveillance,a SHVC region of interest(ROI)encryption algorithm based on face detection is proposed.The ellipse model skin color detection algorithm is used to locate the face region,map to the divided Tile unit,and perform encryption processing on it to cause serious texture distortion on the face.And the algorithm uses restrictive motion estimation and motion prediction processing to prevent drift in the ROI.Finally,the encrypted Tile unit index information is embedded into the Slice segment header extension data to achieve correct detection of the ROI on the decoding side.The algorithm satisfies the encryption requirements of the ROI,has good security and robustness and low complexity,effectively inhibits error diffusion and introduces about 12.54% compression load.