英語(yǔ)翻譯
Residual coefficients are primarily the transform coefficients of the difference between a macroblock and its motion
compensated version,after they have been quantized and entropy coded.Our experiments have shown that these
residual coefficients do carry a certain amount of “residual visual information” that may convey to an adversary the
content of the compressed video if not properly secured.Thus,it becomes necessary to encrypt the residual information.
Figure 2 shows that without having to decoding the reference frame or motion vectors,the luma residue coefficients of a
P frame will provide an adversary a delineation of the original frame.Therefore,we need to protect the residual
coefficients.However,further experiments reveal that not all of the luma and chroma coefficients need to be secured.
Figure 3 demonstrates the effect of scrambling all the residual luma and chroma DC coefficients of a P frame when its
motion vectors and reference frame have not been encrypted.As can be seen,for the most part encrypting the DC
coefficients was sufficient to obscure many of the details in the frame.Any information that did leak through was due to
unsecured motion vectors and reference frame.