[Mp4-tech] [MP4-Video] Mismatch control and Inverse quantization

[pankaj bajpai]

Dear All,
I have a question regarding mismatch control and Inverse quantization.
As can be seen from the below equations, in both cases, the output/conditons
are based upon eveness/oddness and the attempt is to make the things odd.
I would like to know why there is bias towards oddness?
How it serves the idea of mismatch control?
And any mathematical/physical proof is there or is it just
statistical observation (i have found in some posts that
statistically it has been found that some patterns create 
lot of problem, and this way they can be rectified).
I am not able to figure out the exact reason.
Please give some pointers regarding this
With thanks 
Pankaj 
1. Mismatch control:
sum = (F' [v][u])
F[v][u] = F' [v][u] for all u, v except u = v = 7
F[7][7] =  F' [7][7]           if sum is odd
            if sum is even
           F' [7][7] −1        if F' [7][7 ] is odd
           F' [7][7] +1        if F' [7][7 ] is even
2. Inverse quantization
|F(v,u)| = (2*Q(v,u)+1) * q_scale         if q_scale is odd
|F(v,u)| = (2*Q(v,u)+1) * q_scale  -1     if q_scale is even

-- 
_______________________________________________
Surf the Web in a faster, safer and easier way:
Download Opera 9 at http://www.opera.com
Powered by Outblaze