Optimized image segmentation and its effect on classification accuracy

Y. Gao, N. Kerle, J.F. Mas, A. Navarrete, I. Niemeyer

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

2 Downloads (Pure)

Abstract

Image segmentation is a preliminary and critical step in segment-based image analysis. Its proper evaluation ensures that the best segmentation result is used in image classification. In this paper, image segmentations were carried out and the results were evaluated with an objective function that aims at maximizing homogeneity within segments and separability between neighbouring segments. The segmented images were classified with Maximum Likelihood Classifier (MLC) and the classification results were evaluated with independent ground data. The optimal segmentation, i.e. with the highest objective function value, also resulted in the highest classification accuracy, which is 5.92% higher than that obtained by the segmentation with the lower objective function value, and the difference is significant by McNemar’s test with p= 0.05, p is the significance level. This shows that the objective function is indeed an effective way to determine the optimal segmentations to carry out the classifications. Pixel-based MLC was also carried out to compare with the segment-based classification. Besides free of salt-and-pepper effect, the best-segmentationbased classification obtained accuracy 2.3% higher than obtained by the pixel-based classification. Though by McNemar’s test, the difference is not significance, with p=0.05. This result seems to suggest that the benefit of segmentation-based classification lies not only in the segmentation step, which alone leads to marginal classification improvement, but that the use of segments’ shape, contextual as well as spectral information, is needed to increase accuracy significantly.
Original languageEnglish
Title of host publicationProceedings of the 5th International symposium on Spatial Data Quality, SDQ 2007
Subtitle of host publicationModelling qualities in space and time, ITC, Enschede, The Netherlands, 13-15 June, 2007
Place of PublicationEnschede, The Netherlands
PublisherInternational Institute for Geo-Information Science and Earth Observation
Number of pages4
Publication statusPublished - 2007

Fingerprint

segmentation
pixel
effect
image classification
image analysis
homogeneity
salt

Keywords

  • ADLIB-ART-1430
  • ESA

Cite this

Gao, Y., Kerle, N., Mas, J. F., Navarrete, A., & Niemeyer, I. (2007). Optimized image segmentation and its effect on classification accuracy. In Proceedings of the 5th International symposium on Spatial Data Quality, SDQ 2007: Modelling qualities in space and time, ITC, Enschede, The Netherlands, 13-15 June, 2007 Enschede, The Netherlands: International Institute for Geo-Information Science and Earth Observation.
Gao, Y. ; Kerle, N. ; Mas, J.F. ; Navarrete, A. ; Niemeyer, I. / Optimized image segmentation and its effect on classification accuracy. Proceedings of the 5th International symposium on Spatial Data Quality, SDQ 2007: Modelling qualities in space and time, ITC, Enschede, The Netherlands, 13-15 June, 2007. Enschede, The Netherlands : International Institute for Geo-Information Science and Earth Observation, 2007.
@inproceedings{f00ae0a214724f82b60754162a708e14,
title = "Optimized image segmentation and its effect on classification accuracy",
abstract = "Image segmentation is a preliminary and critical step in segment-based image analysis. Its proper evaluation ensures that the best segmentation result is used in image classification. In this paper, image segmentations were carried out and the results were evaluated with an objective function that aims at maximizing homogeneity within segments and separability between neighbouring segments. The segmented images were classified with Maximum Likelihood Classifier (MLC) and the classification results were evaluated with independent ground data. The optimal segmentation, i.e. with the highest objective function value, also resulted in the highest classification accuracy, which is 5.92{\%} higher than that obtained by the segmentation with the lower objective function value, and the difference is significant by McNemar’s test with p= 0.05, p is the significance level. This shows that the objective function is indeed an effective way to determine the optimal segmentations to carry out the classifications. Pixel-based MLC was also carried out to compare with the segment-based classification. Besides free of salt-and-pepper effect, the best-segmentationbased classification obtained accuracy 2.3{\%} higher than obtained by the pixel-based classification. Though by McNemar’s test, the difference is not significance, with p=0.05. This result seems to suggest that the benefit of segmentation-based classification lies not only in the segmentation step, which alone leads to marginal classification improvement, but that the use of segments’ shape, contextual as well as spectral information, is needed to increase accuracy significantly.",
keywords = "ADLIB-ART-1430, ESA",
author = "Y. Gao and N. Kerle and J.F. Mas and A. Navarrete and I. Niemeyer",
year = "2007",
language = "English",
booktitle = "Proceedings of the 5th International symposium on Spatial Data Quality, SDQ 2007",
publisher = "International Institute for Geo-Information Science and Earth Observation",
address = "Netherlands",

}

Gao, Y, Kerle, N, Mas, JF, Navarrete, A & Niemeyer, I 2007, Optimized image segmentation and its effect on classification accuracy. in Proceedings of the 5th International symposium on Spatial Data Quality, SDQ 2007: Modelling qualities in space and time, ITC, Enschede, The Netherlands, 13-15 June, 2007. International Institute for Geo-Information Science and Earth Observation, Enschede, The Netherlands.

Optimized image segmentation and its effect on classification accuracy. / Gao, Y.; Kerle, N.; Mas, J.F.; Navarrete, A.; Niemeyer, I.

Proceedings of the 5th International symposium on Spatial Data Quality, SDQ 2007: Modelling qualities in space and time, ITC, Enschede, The Netherlands, 13-15 June, 2007. Enschede, The Netherlands : International Institute for Geo-Information Science and Earth Observation, 2007.

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Optimized image segmentation and its effect on classification accuracy

AU - Gao, Y.

AU - Kerle, N.

AU - Mas, J.F.

AU - Navarrete, A.

AU - Niemeyer, I.

PY - 2007

Y1 - 2007

N2 - Image segmentation is a preliminary and critical step in segment-based image analysis. Its proper evaluation ensures that the best segmentation result is used in image classification. In this paper, image segmentations were carried out and the results were evaluated with an objective function that aims at maximizing homogeneity within segments and separability between neighbouring segments. The segmented images were classified with Maximum Likelihood Classifier (MLC) and the classification results were evaluated with independent ground data. The optimal segmentation, i.e. with the highest objective function value, also resulted in the highest classification accuracy, which is 5.92% higher than that obtained by the segmentation with the lower objective function value, and the difference is significant by McNemar’s test with p= 0.05, p is the significance level. This shows that the objective function is indeed an effective way to determine the optimal segmentations to carry out the classifications. Pixel-based MLC was also carried out to compare with the segment-based classification. Besides free of salt-and-pepper effect, the best-segmentationbased classification obtained accuracy 2.3% higher than obtained by the pixel-based classification. Though by McNemar’s test, the difference is not significance, with p=0.05. This result seems to suggest that the benefit of segmentation-based classification lies not only in the segmentation step, which alone leads to marginal classification improvement, but that the use of segments’ shape, contextual as well as spectral information, is needed to increase accuracy significantly.

AB - Image segmentation is a preliminary and critical step in segment-based image analysis. Its proper evaluation ensures that the best segmentation result is used in image classification. In this paper, image segmentations were carried out and the results were evaluated with an objective function that aims at maximizing homogeneity within segments and separability between neighbouring segments. The segmented images were classified with Maximum Likelihood Classifier (MLC) and the classification results were evaluated with independent ground data. The optimal segmentation, i.e. with the highest objective function value, also resulted in the highest classification accuracy, which is 5.92% higher than that obtained by the segmentation with the lower objective function value, and the difference is significant by McNemar’s test with p= 0.05, p is the significance level. This shows that the objective function is indeed an effective way to determine the optimal segmentations to carry out the classifications. Pixel-based MLC was also carried out to compare with the segment-based classification. Besides free of salt-and-pepper effect, the best-segmentationbased classification obtained accuracy 2.3% higher than obtained by the pixel-based classification. Though by McNemar’s test, the difference is not significance, with p=0.05. This result seems to suggest that the benefit of segmentation-based classification lies not only in the segmentation step, which alone leads to marginal classification improvement, but that the use of segments’ shape, contextual as well as spectral information, is needed to increase accuracy significantly.

KW - ADLIB-ART-1430

KW - ESA

M3 - Conference contribution

BT - Proceedings of the 5th International symposium on Spatial Data Quality, SDQ 2007

PB - International Institute for Geo-Information Science and Earth Observation

CY - Enschede, The Netherlands

ER -

Gao Y, Kerle N, Mas JF, Navarrete A, Niemeyer I. Optimized image segmentation and its effect on classification accuracy. In Proceedings of the 5th International symposium on Spatial Data Quality, SDQ 2007: Modelling qualities in space and time, ITC, Enschede, The Netherlands, 13-15 June, 2007. Enschede, The Netherlands: International Institute for Geo-Information Science and Earth Observation. 2007