Jayanta Mukhopadhyay

1
COLOR CONSTANCY IN THE COMPRESSED DOMAIN

• Jayanta Mukhopadhyay
• Department of Computer Science Engineering
• Indian Institute of Technology, Kharagpur,
  721302, India
• jay_at_cse.iitkgp.ernet.in
• Sanjit K. Mitra
• Ming Hsieh Dept. of Electrical Engineering
• University of Southern California
• Los Angeles, CA 90089, USA
• skmitra_at_usc.edu

2
Problem of Color Constancy

• Three factors of image formation
• Objects present in the scene.
• Spectral Energy of Light Sources.
• Spectral Sensitivity of sensors.

3
Same Scene Captured under Different Illumination
Can we transfer colors from one illumination to
another one?
4
Computation of Color Constancy

• Deriving an illumination independent
  representation.
• - Estimation of SPD of Light Source.
• Color Correction
• - Diagonal Correction.

E(?)
ltR, G, Bgt
To perform this computation with DCT coefficients.
5
Different Spatial Domain Approaches

• Gray World Assumption (Buchsbaum (1980), Gershon
  et al. (1988))
• ltR, G, Bgt ltRavg, Gavg, Bavggt
• White World Assumption (Land (1977))
• ltR, G, Bgt ltRmax, Gmax, Bmaxgt

6
Select from a set of Canonical Illuminants

• Observe distribution of points in 2-D
• Chromatic Space.
• Assign SPD of the nearest illuminant.
• Gamut Mapping Approach (Forsyth (1990), Finlayson
  (1996))
• - Existence of chromatic points.
• Color by Correlation (Finlayson et. al. (2001))
• - Relative strength over the
  distribution.
• Nearest Neighbor Approach (Proposed)
• - Mean and Covariance Matrix.
• - Use of Mahalanobis Distance.

7
Processing in the Compressed Domain

• Consists of non-overlapping DCT blocks (of 8 x
  8).
• Use DC coefficients of each block.
• The color space used is Y-Cb-Cr instead of RGB.
• Chromatic Space for Statistical Techniques is the
  Cb-Cr space.

8
Different Algorithms under consideration
9
List of Illuminants
10
Images Captured at Different Illumination
Source http//www.cs.sfu.ca/ colour/data.
11
Performance Metrics

• Estimated SPD EltRE,GE,BEgt
• True SPD T ltRT,GT,BTgt

12
Average ??
13
Average ?rg
14
Average ?RGB
15
Average ?L
16
Average Performance over the canonical set.
17
Average ??
18
Average ?rg
19
Average ?RGB
20
Average ?L
21
Time and Storage Complexities

• nl number of illuminants.
• nc size of the 2-D chromaticity space
• n number of image pixels
• f Fraction of chromaticity space covered.
• aMbA ? a number of Multiplications and b number
  of Additions.

22
Time and Storage Complexities
23
Equivalent No. of Additions per pixel (1 M 3 A)
n512, nc32, nl12, f1
24
Color Correction An Example
Image captured with (solux-4100)
Target Ref. Image (syl-50mr16q)
COR-DCT
MXW-DCT-Y
COR
25
Color Restoration
Original
Enhanced w/o Color Correction
Enhanced with Color Correction
26
Conclusion

• Color-constancy computation in the compressed
  domain
• - requires less time and storage.
• - comparable quality of results.
• Both NN and NN-DCT perform well compared to other
  existing statistical approaches.
• Color constancy computation is useful in
  restoration of colors.

27
Thanks!