Page 216 - Video Coding for Mobile Communications Efficiency, Complexity, and Resilience

Section 8.5. Simulation Results 193

QSIF Foreman @ 8.3333 f.p.s., QP = 10 5 QSIF Foreman @ 8.3333 f.p.s., QP = 10
32.5 10
FS
SMS
32 NSS
OTS
31.5
4
31 10
PSNR Y (dB) 30.5 30 Searched locations
29.5 10 3
29
FS
SMS
28.5
NSS
OTS
2
28 10
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Frame Frame
(a) Prediction quality (b) Computational complexity
4 QSIF Foreman @ 8.3333 f.p.s., QP = 10 QSIF Foreman @ 8.3333 f.p.s., QP = 10
x 10
2 1800
FS FS
SMS SMS
1.8 NSS 1600 NSS
OTS OTS
1.6
1400
1.4
1200
DFD bits 1.2 1 Motion bits 1000
800
0.8
600
0.6
400
0.4
0.2 200
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Frame Frame
(c) DFD bits (d) Motion bits
4
x 10 QSIF Foreman @ 8.3333 f.p.s., QP = 10
2.2
FS
SMS
2
NSS
OTS
1.8
1.6
1.4
Total bits 1.2
1
0.8
0.6
0.4
0.2
0 10 20 30 40 50 60 70 80 90 100
Frame
(e) Total bits
Figure 8.9: Comparison between di+erent block-matching algorithms when applied to QSIF
FOREMAN at 8:33 frames=s and QP = 10