Page 261 - Video Coding for Mobile Communications Efficiency, Complexity, and Resilience

238 Chapter 10. Error Concealment Using Motion Field Interpolation

0.5 1
0.4 0.9
1−w(x n ,y n ) 0.3 w(x n ,y n ) 0.8
0.7
0.2
0.1 0.6
0 0.5
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.8 1 0.2 0.8 1
0.4 0.6 0.4 0.6
y n 0 0.2 y n 0 0.2
0 0
x n x n
(a) BM weights, (1 − ), with =1 (b) MFI weights, , with =1
1 1
0.9 0.9
0.8 0.8
w(x n ,y n ) 0.7 w(x n ,y n ) 0.7
0.6 0.6
0.5 0.5
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.8 1 0.2 0.8 1
0.6 0.6
0 0.4 0 0.4
0 0
y n 0.2 y n 0.2
x n x n
(c) MFI weights, , with =5 (d) MFI weights, , with =50
Figure 10.2: Multihypothesis weights w (·; ·) with di,erent values of the smoothness parameter
Table 10.2: Comparison between conventional temporal concealment and the MFI and BM-MFI
techniques
Conventional temporal
concealment MFI BM-MFI

Displacement One displacement per One displacement per Two displacements
estimation blockusing AV, pel using MFI per pel: one
TR, BM, etc. (weighted sum of produced by MFI
four neighboring and another
vectors) produced by BM
Displacement Translational, same Translational, but on a Multihypothesis
compensa- displacement for pel-by-pel basis motion
tion the whole block. compensation
(each pel is a
weighted sum of
two concealments)