This is very simple to protect this. There are many problems in achieving it. For video streams: Obviously Motion Vector is important than DCT CoeffICents, how is the former better protection? Initial Idea: Directly separated in the video stream, after INTERLEAVE, add different FEC. The problem is how to separate, although the MEPG2 can indeed separate, but in order to separate the reassembly, how much OverLoad is to increase? It seems that it is not possible. After reading the MEPG4 encoding method, we will give up completely. . . The auxiliary information is directly smoke, packaged, passing twice, how? Similarly, Reduntancy is too large, and all the auxiliary information seems to account for more than 1/3 of all BIT streams. . . . It is not not available for both existing standards. Suppose we can share the auxiliary information and DCT coefficients very little OverLoad, how? The result is: This is just a tradeoff sometimes lost the possibility of space Recovery. The result of considering is a crying. I am still too much. Sigh
