LOW COMPUTATIONAL OVERHEAD VIDEO ENCRYPTION FOR WIRELESS MULTIMEDIA DEVICES

Abstract

Selective encryption is a promising technique which provides privacy protection for multimedia technology. In this thesis we propose a scheme to optimize computational cost for energy critical wireless multimedia applications. Latest research in selective encryption algorithm proposed to encrypt syntax elements such as intra prediction modes, sign bit of residual coefficients, along with sign bit of motion vectors. Such syntax elements are sensitive enough to provide effective scrambling effect with tractable computational cost. The proposed algorithm selects code words for encryption based on scene transitions in compressed frames. The ratio of intracoded macroblocks in inter frames is calculated and compared with a threshold value to detect scene transitions. Further, Based on statistical properties of the video frame, a dynamic threshold model for scene change detection has been proposed. In case of video frames with scene transition, intra modes and sign bit of residuals are chosen as code words to encrypt, whereas in case of no scene transitions, sign bit of motion vectors are chosen as sensitive code words for encryption. Experimental results showed that the proposed algorithm maintains guaranteed security with low computational overhead.