MPEG-2 Video Coding - Digital Video and Audio Broadcasting Technology - page 125

relatively symmetric to the diagonal axis from top left to bottom right. The
matrix is, therefore, read out in a zig-zag scanning process which then pro-
vides a large number of adjacent zeroes. These can then be variable-length
coded in the next step, resulting in a very large data reduction. The quanti-
zation is the only 'adjusting screw' for controlling the data rate of the
video elementary stream.

8 16192226272934
16 16 22 24 27 29 34 37

19 22 26 27 29 34 34 38

22 22 26 27 29 34 37 40

22 26 27 29 32 35 40 48

26 27 29 32 35 40 48 58

26 27 29 34 38 46 56 69

27 29 35 38 46 56 69 83

Q(v,u)

scale_factor = 2 ;

QF(v,u) = F(v,u) / Q(v,u) / scale_factor ;

Fig. 7.20. Quantization after the DCT

With 4:2:0, four 8 x 8 Y pixel blocks and one 8 x 8 C B and 8 x 8 C R
pixel block each are combined to form one macroblock (Fig.7.20.). The
quantization for Y, C B and C R can be changed by means of a special quan-
tizer scale factor from macroblock to macroblock. This factor alters all
quantization factors either of the standard MPEG tables or of the quantiza-
tion tables provided by the encoder, by a simple multiplication by a certain
factor. The complete quantization table can only be exchanged at sequence
level at certain times, as will be seen later.
This transform coding followed by quantization must be performed for
the Y pixel plane and for the C B and C R planes.