Video Tone Mapping dr. Francesco Banterle - - PowerPoint PPT Presentation
Video Tone Mapping dr. Francesco Banterle francesco.banterle@isti.cnr.it Video Tone Mapping How do HDR videos behave when applying a TMO for each frame? video sequence in this presentation from http://www.hdrv.org by Jonas Unger Video Tone
francesco.banterle@isti.cnr.it
for each frame?
video sequence in this presentation from http://www.hdrv.org by Jonas Unger
Sigmoid TMO
Sigmoid TMO
Adaptive Logarithmic TMO
Adaptive Logarithmic TMO
temporal flicker
Frame t Frame t + 1
50 100 150 200 250 300 350 10 10
1
10
2
10
3
10
4
10
5
Frames Luminance cd/m2
geometric mean value max value mean value min value
50 100 150 200 250 300 350 10 10
1
10
2
10
3
10
4
10
5
Frames Luminance cd/m2
geometric mean value max value mean value min value
50 100 150 200 250 300 350 10 10
1
10
2
10
3
10
4
10
5
Frames Luminance cd/m2
geometric mean value max value mean value min value
50 100 150 200 250 300 350 10 10
1
10
2
10
3
10
4
10
5
Frames Luminance cd/m2
geometric mean value max value mean value min value
50 100 150 200 250 300 350 10 10
1
10
2
10
3
10
4
10
5
Frames Luminance cd/m2
geometric mean value max value mean value min value
50 100 150 200 250 300 350 10 10
1
10
2
10
3
10
4
10
5
Frames Luminance cd/m2
geometric mean value max value mean value min value
pass filter: box, Gaussian, etc.
the temporal domain!
50 100 150 200 250 300 350 10
3
10
4
10
5
Frames Luminance cd/m2
Smoothed signal Original signal
[Kiser et al. 2012]
a = 0.18 × 22 B−A
A+B
A = Lw,max − Lw B = Lw − Lw,min Ld(x) = Lm(x) ✓ 1 + Lm(x)
L2
white
◆ 1 + Lm(x) Lm(x) = a Lw Lw(x)
At = (1 − αA)At−1 + αAA αA ∈ [0, 1] Bt = (1 − αB)Bt−1 + αBB αB ∈ [0, 1] at = (1 − αa)at−1 + αaa αa ∈ [0, 1] Smoothing for each derived statistic:
frames of a continuous cut [Kang et al. 2003]
50 100 150 200 250 300 350 10
3
10
4
10
5
Frames Luminance cd/m2
Smoothed signal Original signal
Tone mapping frame 259
50 100 150 200 250 300 350 10
3
10
4
10
5
Frames Luminance cd/m2
Smoothed signal Original signal
Tone mapping frame 259
50 100 150 200 250 300 350 10
3
10
4
10
5
Frames Luminance cd/m2
Smoothed signal Original signal
Tone mapping frame 259
Frame 259
Frame 259
general solution
ensured [Boitard et al. 2012]
L0
d(x) = Ld(x)Lw × Ld,max
Lw,max × Ld Lw Lw,max = Ld Ld,max
Lw(x, y, t) = Lw(x + u, y + v, t + 1) C = X
x
X
y
✓ Ld(x, y, t) − Ld(x + u, y + v, t + 1) ◆2
HDR
t t+1
HDR
(u,v)
LDR LDR
(u,v)
filter [Aydin et al. 2015]