24
Evolving Color Constancy Marc Ebner Universität Würzburg, Germany Pattern Recognition Letters 27 2006 1220- 1229 Elsevier

Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Embed Size (px)

Citation preview

Page 1: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Evolving Color Constancy

Marc EbnerUniversität Würzburg, GermanyPattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Page 2: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Algorithms for color constancy Gamut – constraint methods Perspective color constancy Color by correlation The gray world assumption Recovery of basis function coefficients Mechanisms of light adaptation coupled with movements Neural networks Comprehensive color normalization Committee – based methods Algorithms based on the dichromatic color model Computation of intrinsic images

Page 3: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

PE ( Articial Retina ) PE : a rectangular grid of processing ele

ments Better than neural nets, quite complicate

d.

Page 4: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Processing elements 1 PE for 1 image pixel 3 layers of PEs carrying out results on the

3 image bands red, green, and blue. : Estimate of the illuminant ( color

of input pixel ) The data from other neighboring PEs Initially, ( : pixel value )

Page 5: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier
Page 6: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Conclusion

Only the current color channel ( band ) is used.

Average data from neighboring elements.

Page 7: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Parallel algorithm

The gray world assumption The reflectance , : distributed over the interval [0,1] From PE, N : the number of image pixels.

Page 8: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Parallel algorithm a ( x, y ): an estimate of local space a

verage color for each image pixel N ( x, y ): a set of neighboring elemen

ts

( 1 ) Average the data ( 2 ) Slowly add the color of the curren

t pixel ( p : small percentage )

Page 9: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Parallel algorithm The two equations, ( 1 )&( 2 ) ,are carried out until conv

ergence.

1000, 2000, 3000, 4000, 5000

Page 10: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier
Page 11: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Local space average color 1 50 200 1000

The parallel algorithm 1000

Page 12: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Reference

Ebner, M., 2001. Evolving color constancy for an artificial retina. Genetic Programming: Proc. of the 4thEuropean Conference, EuroGP 2001, Lake Como, Italy. Springer-Verlag, Berlin, pp. 11–22.

Ebner, M., 2004. A parallel algorithm for color constancy. J. Parallel Distributed Comput. 64 (1), 79–88.

Page 13: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Why Mondrian has been chosen First introduced by Edwin Land No curve and angle. No shade and textur

e. Neither uniformly colored nor uniformly

bright. Resemble better the more colorful work

of Klee or Lohse. Anya Hurlbert, 1999

Page 14: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Paul Klee 南方突尼西亞人花園 Tunisian Gardens

1919

Ref. www.writedesignonline.com/history-culture/bauhaus.htm

Page 15: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Richard Paul Lohse Thematic series in 18 colours A, 1982

Squares formed by colour groups 1944/2

Ref. www.lohse.ch/bio_e.html

Page 16: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Mondrian Piet Mondrian, Composition A, 1923

www.cartage.org.lb/en/themes/Arts/painting/20th-century/art-sake/artsake.htm

Page 17: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Typical Mondrian stimuli Yellowish daylight ; bluish daylight

2 grey papers ( third from the top on the left )

Page 18: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

The experiment of Kraft and Brainard Look through a window into a box A grey test surface against the back wall A Mondrian-like panel A tube wrapped in tin foil A cube, pyramid and tube made from grey cardboard

Page 19: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Local surround

Neutral-illuminant ; Orange-red

Page 20: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Spatial Mean

Neutral-illuminant ; pale-red

Page 21: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Maximum Flux

Neutral-illuminant ; yellow-illuminant

Page 22: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Results

Color constancy

Page 23: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Anya Hurlbert, 2007 Unknown why humans need color consta

ncy. Color? Shape? How is color constancy measured? with d

ifficulty. Mondrians? How is color constancy achieved? More t

han one mechanism. Color processing in the brain.

Retinex

Page 24: Evolving Color Constancy Marc Ebner Universit ä t W ü rzburg, Germany Pattern Recognition Letters 27 ( 2006 ) 1220-1229 Elsevier

Reference Hurlbert A (1999) Colour vision: is colour

constancy real? Current Biology 9:R558–R561.

Hurlbert, A. (2007). Colour constancy. Current Biology, 17(21), R906-7.

JM Kraft and DH Brainard, Mechanisms of color constancy under nearly natural viewing. Proc Natl Acad Sci USA 96 (1999), pp. 307–312.