Publication detail

Retrospective Illumination Correction of Retinal Images

Kubecka L, Kolar R, Jan J, Jirik R

Original Title

Retrospective Illumination Correction of Retinal Images

Czech Title

Retrospektivní korekce osvětlení obrazů sítnice

English Title

Retrospective Illumination Correction of Retinal Images

Type

conference paper

Language

en

Original Abstract

A method for correction of non-homogenous illumination based on optimization of parameters of B-spline shading model in respect to the Shannons entropy is presented. An analytical gradient term of the entropy criterion has been derived allowing to use the efficient L-BFGS optimization algorithm. The method has been tested on the set of 20 retinal images and has shown ability to substantially eliminate the illumination distortion. The application of this method for reliable image segmentation or registration was tested.

Czech abstract

Je uvedena metoda korekce nehomogenního osvětlení založená na optimalizaci modelu osvětlení založeném na Shanonnově entropii. Byl nalezen analytický tvar pro parciální derivace potřebné v optimalizačním kroku. Metoda byla testována na retinálních obrazech. m

English abstract

A method for correction of non-homogenous illumination based on optimization of parameters of B-spline shading model in respect to the Shannons entropy is presented. An analytical gradient term of the entropy criterion has been derived allowing to use the efficient L-BFGS optimization algorithm. The method has been tested on the set of 20 retinal images and has shown ability to substantially eliminate the illumination distortion. The application of this method for reliable image segmentation or registration was tested.

Keywords

illumination correction, optimization, Parzen windowing, retina

RIV year

2006

Released

26.06.2006

Location

Brno

Pages from

257

Pages to

515

Pages count

259

BibTex


@inproceedings{BUT18968,
  author="Libor {Kubečka} and Jiří {Jan} and Radim {Kolář} and Radovan {Jiřík}",
  title="Retrospective Illumination Correction of Retinal Images",
  annote="A method for correction of non-homogenous illumination based on optimization of
parameters of B-spline shading model in respect to the Shannons entropy is presented. An
analytical gradient term of the entropy criterion has been derived allowing to use the efficient
L-BFGS optimization algorithm. The method has been tested on the set of 20 retinal images
and has shown ability to substantially eliminate the illumination distortion. The application
of this method for reliable image segmentation or registration was tested.",
  booktitle="Proceedings of 17th International EURASIP Conference BIOSIGNAL 2004",
  chapter="18968",
  year="2006",
  month="june",
  pages="257",
  type="conference paper"
}