收录:
摘要:
In this paper, a content-based encryption scheme for wireless H.264 video in the compressed domain is proposed. First, the information of prediction mode, motion vector and prediction residual energy is extracted from H.264 bitstream and then jointly utilized to compute measures to classify the complexity of the video content in terms of texture and motion. Then according to the classification results, videos with different content features are encrypted with selective encryption in different ways in the unit of GOP (GOP, Group of Pictures). In the case of videos of complex texture, we encrypt all the intra prediction modes and DCT coefficients in the GOP; as for videos of not that much texture, the number of high-band coefficients to encrypt will be reduced. As for videos of high motion intensity, we encrypt MVDs in all the P frames in the GOP; while as for those of medium or low intensity, only MVDs in parts of P frames in the GOP are to encrypt. Experiments have shown that compared to the existing algorithm of SEH264, the proposed content-based video encryption scheme can provide good performance in both aspects of security as well as computational complexity. Moreover, the proposed encryption scheme is implemented entirely in the compressed domain and format-compliant with the H.264 bitstream without influence on the compression ratio. © 2012 IEEE.
关键词:
通讯作者信息:
电子邮件地址: