Chromatic Framework for Vision in Bad Weather

1
Chromatic Framework for Vision in Bad
Weather
Srinivasa G. Narasimhan and Shree K.
Nayar Computer Science Department Columbia
University IEEE CVPR Conference June
2000, Hilton Head Island, USA Sponsors ONR
MURI , NSF
2
The Colors of Bad Weather
Clear Day
B
R
G
3
Prior Work

• Overviews Middleton 1952 , McCartney 1976
  
• Haze Hulburt 1946 , Hidy 1972
• Fog Koshmeider 1924 , George 1951 , Myers
  1968

• Vision
• Cozman Krotkov 1997 - Depth
  Cues from Iso-Intensities
• Nayar Narasimhan 1999 -
  Complete Structure Restricted weather
  conditions

4
Direct Transmission and Airlight Models
Object
Observer
d
5
Dichromatic Atmospheric Scattering Model
( Nayar Narasimhan, 1999 )
B
E
R
G
6
Dichromatic Planes
Direct Transmission Color
Dichromatic Plane
Airlight Color
7
Direction of Airlight ( Fog or Haze )
Color
Plane 1 (Scene Point X)
Weather Condition 1
Weather Condition 2
8
Depth from Unknown Weather Conditions
Scattering Coefficients
( Unknown )
9
Direct Transmission Ratio
Direct Transmission Color
Dichromatic Plane
Airlight Color
10
Sky Brightnesses
Direct Transmission Color
Dichromatic Plane
Airlight Color
11
Results with a Synthetic Scene
Color Patches
Rotated Structure
12
Simulation Results
Noise
0
0.5
1.0
1.5
2.0
2.5
3.0
Estimated
100
100.02
100.55
100.65
101.26
103.23
104.84
258.2
260.13
Estimated
255
255.02
256.61
263.45
255.4
Depth Error ()
0.0
0.82
0.42
0.58
0.89
0.96
0.76
Actual Values

13
Structure from Two Weather Conditions
Scene under two different Hazy Conditions
14
Structure from Two Weather Conditions
Scene under two different Foggy Conditions
15
True Color Recovery - Color Cube Boundary
Algorithm
1
2
3
B
O
G
R
16
True Color Recovery
Scene under two different Foggy Conditions
( Brightened )
17
Summary

• Airlight Color from Dichromatic Planes

• Scene Depth from Dichromatic Constraints

• True Color from Color Boundary Constraint