基于目前的DCT系数加密算法在安全性、压缩比和信噪比上都各自存在缺点，提出了一种新的适用于H264/AVC的DCT系数分层置乱算法(DCLSA)。该算法针对H264/AVC中4×4 DCT变换的特点，首先将同一宏块中每个4×4块DCT系数进行块间分层，构建系数分层模型，然后可根据安全性要求对不同层系数进行随机置乱，实现加密编码。通过性能比较和具体实验效果分析，此算法具有更高的安全性、更优的压缩比和较好的信噪比，适合于H.264/AVC的网络安全应用。

This paper presents a novel DCT coefficients layered scrambling algorithm (DCLSA) for H.264/AVC based on the deficiencies of current DCT coefficients encryption algorithms in respect of security,compact ratio and signal to noise ratio. According to the characteristics of the 4×4 DCT transform of H.264/AVC,the algorithm first divides the coefficients of every 4×4 block of the same macro block into several layers,and build a coefficients layered model. Then different layer is scrambled,respectively,by security requirements to achieve secure video coding. DCLSA has shown significant advantages on security,compact ratio and signal to noise ratio through performance comparisons with other algorithms and concrete experimental results,thus making it especially suitable for secure network applications.