robleskelly-radiancefunctionestimation-2004.pdf (195.76 kB)
Radiance function estimation for object classification
conference contribution
posted on 2004-01-01, 00:00 authored by Antonio Robles-KellyAntonio Robles-Kelly, Edwin R HancockThis paper describes a simple method for estimating the surface radiance function from single images of smooth surfaces made of materials whose reflectance function is isotropic and monotonic. The method makes use of an implicit mapping of the Gauss map between the surface and a unit sphere. By assuming the material brightness is monotonic with respect to the angle between the illuminant direction and the surface normal, we show how the radiance function can be represented by a polar function on the unit sphere. Under conditions in which the light source direction and the viewer direction are identical, we show how the recovery of the radiance function may be posed as that of estimating a tabular representation of this polar function. A simple differential geometry analysis shows how the tabular representation of the radiance function can be obtained using the cumulative distribution of image gradients. We illustrate the utility of the tabular representation of the radiance function for purposes of material classification.
History
Event
CIARP 2004 Pattern Recognition. Iberoamerican Congress (9th : 2004 : Puebla, Mexico)Volume
3287Series
Lecture Notes in Computer Science Book SeriesPagination
67 - 75Publisher
Springer-Verlag BerlinLocation
Puebla, MEXICOPlace of publication
Heidelberg, GermanyPublisher DOI
Start date
2004-10-26End date
2004-10-29ISSN
0302-9743eISSN
1611-3349ISBN-10
3-540-23527-2Language
engPublication classification
E1.1 Full written paper - refereedCopyright notice
2004, Springer-Verlag Berlin HeidelbergEditor/Contributor(s)
Alberto Sanfeliu, José Trinidad, Jesús OchoaTitle of proceedings
CIARP2004 : Proceedings of the 9th Iberoamerican Congress on Pattern RecognitionUsage metrics
Categories
No categories selectedKeywords
Licence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC