A contextual classifier that only requires one prototype pixel for each class

Publication: Research - peer-reviewArticle in proceedings – Annual report year: 2001

Standard

A contextual classifier that only requires one prototype pixel for each class. / Maletti, Gabriela Mariel; Ersbøll, Bjarne Kjær; Conradsen, Knut.

Proceedings on IEEE Nuclear Science Symposium Conference Record. Vol. 3 2001. p. 1385-1389.

Publication: Research - peer-reviewArticle in proceedings – Annual report year: 2001

Harvard

Maletti, GM, Ersbøll, BK & Conradsen, K 2001, 'A contextual classifier that only requires one prototype pixel for each class'. in Proceedings on IEEE Nuclear Science Symposium Conference Record. vol. 3, pp. 1385-1389.

APA

Maletti, G. M., Ersbøll, B. K., & Conradsen, K. (2001). A contextual classifier that only requires one prototype pixel for each class. In Proceedings on IEEE Nuclear Science Symposium Conference Record. (Vol. 3, pp. 1385-1389)

CBE

Maletti GM, Ersbøll BK, Conradsen K. 2001. A contextual classifier that only requires one prototype pixel for each class. In Proceedings on IEEE Nuclear Science Symposium Conference Record. pp. 1385-1389.

MLA

Vancouver

Maletti GM, Ersbøll BK, Conradsen K. A contextual classifier that only requires one prototype pixel for each class. In Proceedings on IEEE Nuclear Science Symposium Conference Record. Vol. 3. 2001. p. 1385-1389.

Author

Maletti, Gabriela Mariel; Ersbøll, Bjarne Kjær; Conradsen, Knut / A contextual classifier that only requires one prototype pixel for each class.

Proceedings on IEEE Nuclear Science Symposium Conference Record. Vol. 3 2001. p. 1385-1389.

Publication: Research - peer-reviewArticle in proceedings – Annual report year: 2001

Bibtex

@inbook{f82df4a3e2ff44e8a8e1170194850144,
title = "A contextual classifier that only requires one prototype pixel for each class",
author = "Maletti, {Gabriela Mariel} and Ersbøll, {Bjarne Kjær} and Knut Conradsen",
year = "2001",
volume = "3",
isbn = "0-7803-7324-3",
pages = "1385-1389",
booktitle = "Proceedings on IEEE Nuclear Science Symposium Conference Record",

}

RIS

TY - GEN

T1 - A contextual classifier that only requires one prototype pixel for each class

A1 - Maletti,Gabriela Mariel

A1 - Ersbøll,Bjarne Kjær

A1 - Conradsen,Knut

AU - Maletti,Gabriela Mariel

AU - Ersbøll,Bjarne Kjær

AU - Conradsen,Knut

PY - 2001

Y1 - 2001

N2 - A three stage scheme for classification of multi-spectral images is proposed. In each stage, statistics of each class present in the image are estimated. The user is required to provide only one prototype pixel for each class to be seeded into a homogeneous region. The algorithm starts by generating optimum initial training sets, one for each class, maximizing the redundancy in the data sets. These sets are the realizations of the maximal discs centered on the prototype pixels for which it is true that all the elements belong to the same class as the center one. Afterwards a region growing algorithm increases the sample size providing more statistically valid samples of the classes. Final classification of each pixel is done by comparison of the statistical behavior of the neighborhood of each pixel with the statistical behavior of the classes. A critical sample size obtained from a model constructed with experimental data is used in this stage. The algorithm was tested with the Kappa coefficient k on synthetical images and compared with K-means (k~=0.41) and a similar scheme that uses spectral means (k~=0.75) instead of histograms (k~=0.90). Results are shown on a dermatological image with a malignant melanoma.

AB - A three stage scheme for classification of multi-spectral images is proposed. In each stage, statistics of each class present in the image are estimated. The user is required to provide only one prototype pixel for each class to be seeded into a homogeneous region. The algorithm starts by generating optimum initial training sets, one for each class, maximizing the redundancy in the data sets. These sets are the realizations of the maximal discs centered on the prototype pixels for which it is true that all the elements belong to the same class as the center one. Afterwards a region growing algorithm increases the sample size providing more statistically valid samples of the classes. Final classification of each pixel is done by comparison of the statistical behavior of the neighborhood of each pixel with the statistical behavior of the classes. A critical sample size obtained from a model constructed with experimental data is used in this stage. The algorithm was tested with the Kappa coefficient k on synthetical images and compared with K-means (k~=0.41) and a similar scheme that uses spectral means (k~=0.75) instead of histograms (k~=0.90). Results are shown on a dermatological image with a malignant melanoma.

SN - 0-7803-7324-3

VL - 3

BT - Proceedings on IEEE Nuclear Science Symposium Conference Record

T2 - Proceedings on IEEE Nuclear Science Symposium Conference Record

SP - 1385

EP - 1389

ER -