Trevor Owens and Kate Saenko and Ayan Chakrabarti and Ying Xiong and Todd Zickler and Trevor Darrell

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2011-23

April 4, 2011

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2011/EECS-2011-23.pdf

Color constancy is the ability to infer stable material colors despite changes in lighting, and it is typically ad- dressed computationally using a single image as input. In many recognition and retrieval applications, we have ac- cess to image sets that contain multiple views of the same object in different environments; we show in this technical report and a related publication [8], that correspondences between these images provide important constraints that can improve color constancy. In this report, we present an- other method to solve the multi-view color constancy prob- lem, the Ratio Method. This method provides a means to recover estimates of underlying surface reflectance based on joint estimation of these surface properties and the illu- minants present in multiple images. In contrast to the multi- view Spatial Correlations method (MVSC), this method can leverage any single image color constancy method as a bootstrap for the multi-view solution. The method ex- ploits image correspondences obtained by various align- ment techniques, and we show examples based on match- ing local region features. Our results show that the Ra- tio Method performs similarly to the MVSC method, both of which are improvements over a baseline single-view method.

---

BibTeX citation:

@techreport{Owens:EECS-2011-23,
    Author= {Owens, Trevor and Saenko, Kate and Chakrabarti, Ayan and Xiong, Ying and Zickler, Todd and Darrell, Trevor},
    Title= {The Ratio Method for Multi-view Color Constancy},
    Year= {2011},
    Month= {Apr},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2011/EECS-2011-23.html},
    Number= {UCB/EECS-2011-23},
    Abstract= {Color constancy is the ability to infer stable material colors despite changes in lighting, and it is typically ad- dressed computationally using a single image as input. In many recognition and retrieval applications, we have ac- cess to image sets that contain multiple views of the same object in different environments; we show in this technical report and a related publication [8], that correspondences between these images provide important constraints that can improve color constancy. In this report, we present an- other method to solve the multi-view color constancy prob- lem, the Ratio Method. This method provides a means to recover estimates of underlying surface reflectance based on joint estimation of these surface properties and the illu- minants present in multiple images. In contrast to the multi- view Spatial Correlations method (MVSC), this method can leverage any single image color constancy method as a bootstrap for the multi-view solution. The method ex- ploits image correspondences obtained by various align- ment techniques, and we show examples based on match- ing local region features. Our results show that the Ra- tio Method performs similarly to the MVSC method, both of which are improvements over a baseline single-view method.},
}

EndNote citation:

%0 Report
%A Owens, Trevor 
%A Saenko, Kate 
%A Chakrabarti, Ayan 
%A Xiong, Ying 
%A Zickler, Todd 
%A Darrell, Trevor 
%T The Ratio Method for Multi-view Color Constancy
%I EECS Department, University of California, Berkeley
%D 2011
%8 April 4
%@ UCB/EECS-2011-23
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2011/EECS-2011-23.html
%F Owens:EECS-2011-23