08.01.2013 Views

DigitalVideoAndHDTVAlgorithmsAndInterfaces.pdf

DigitalVideoAndHDTVAlgorithmsAndInterfaces.pdf

DigitalVideoAndHDTVAlgorithmsAndInterfaces.pdf

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Table 40.7 MPEG<br />

macroblock types Prediction<br />

Typ. quantizer<br />

matrix<br />

Intra None – the macroblock is self-contained Perceptual<br />

Nonintra<br />

Backward<br />

predictive-coded<br />

Forward<br />

predictive-coded<br />

Bidirectionally<br />

predictive-coded<br />

Table 40.8 MPEG<br />

picture coding types<br />

Binary<br />

code<br />

I-picture 001 Yes Intra<br />

Predicts from a future anchor picture Flat<br />

Predicts from a past anchor picture Flat<br />

Averages predictions from past and future<br />

anchor pictures<br />

Anchor<br />

picture? Permitted macroblock types<br />

P-picture 010 Yes Intra<br />

Forward predictive-coded<br />

B-picture 011 No Intra<br />

Forward predictive-coded<br />

Backward predictive-coded<br />

Bidirectionally predictive-coded<br />

Table 40.9 MPEG-2<br />

prediction modes<br />

Max. MVs<br />

for Description back. fwd.<br />

Frame prediction (P, B)-frames Predictions are made for the frame, using<br />

data from one or two previously<br />

reconstructed frames.<br />

Field prediction (P, B)-frames,<br />

(P, B)-fields<br />

16×8 motion<br />

compensation<br />

(16×8 MC)<br />

Dual prime P-fields with<br />

no intervening<br />

B-pictures<br />

Dual prime P-frames with<br />

no intervening<br />

B-pictures<br />

Predictions are made independently for<br />

each field, using data from one or two<br />

previously reconstructed fields.<br />

(P, B)-fields The upper 16×8 and lower 16×8 regions<br />

of the macroblock are predicted separately.<br />

(This is completely unrelated to top and<br />

bottom fields.)<br />

CHAPTER 40 MPEG-2 VIDEO COMPRESSION 483<br />

Flat<br />

Two motion vectors are derived from the<br />

transmitted vector and a small differential<br />

motion vector (DMV, -1, 0, or +1); these<br />

are used to form predictions from two<br />

anchor fields (one top, one bottom), which<br />

are averaged to form the predictor.<br />

As in dual prime for P-fields (above), but<br />

repeated for 2 fields; 4 predictions are<br />

made and averaged.<br />

1 1<br />

1 1<br />

2 2<br />

1 1<br />

1 1

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!