Object-based random forest classification for mapping floodplain vegetation structure from nation-wide CIR and LiDAR datasets

L. Kooistra, E.T. Kuilder, C.A. Mucher

Research output: Chapter in Book/Report/Conference proceedingConference paper

2 Citations (Scopus)

Abstract

Very high resolution aerial images and LiDAR (AHN2) datasets with a national coverage provide opportunities to produce vegetation maps automatically. As such the entire area of the river floodplains in the Netherlands may be mapped with high accuracy and regular updates, capturing the dynamic state of the vegetation. In this study, these fused datasets are used to map the vegetation of 936 ha of the floodplain on the north-side of the river Nederrijn near Wageningen into ten vegetation structure classes. The method follows object-based image analysis principles. Objects are defined in segmentation and subsequently labeled using the ensemble-tree classifier random forest. The mapping scale is controlled by selecting segmentation parameters from quantified discrepancies between reference polygons and segmented objects. Effects on the mapping scale of different reference polygons and different segmentation data is investigated. The results show that it is important to be able to select the right segmentation parameters to control the mapping scale. A discrepancy measure with reference polygons is a suitable method to do this objectively. The use of random forest classification on the objects resulted in an estimated classification accuracy of 86% on the basis of the built-in cross-validation estimate of random forest. Variable importance measures of random forest showed that the AHN2 lidar dataset is a valuable addition to the spectral information contained in the aerial images in the classification.

Original languageEnglish
Title of host publication6th Workshop on Hyperspectral Image and Signal Processing
Subtitle of host publicationEvolution in Remote Sensing, WHISPERS 2014
PublisherIEEE computer society
Number of pages4
ISBN (Electronic)9781467390125
ISBN (Print)9781467390132
DOIs
Publication statusPublished - 26 Oct 2017
Event6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014 - Lausanne, Switzerland
Duration: 24 Jun 201427 Jun 2014

Conference

Conference6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014
CountrySwitzerland
CityLausanne
Period24/06/1427/06/14

Fingerprint

Rivers
Antennas
Optical radar
Image analysis
Classifiers

Keywords

  • Object Based Image Analysis (OBIA)
  • reference polygons
  • segmentation optimization
  • variable importance
  • vegetation structure classes

Cite this

Kooistra, L., Kuilder, E. T., & Mucher, C. A. (2017). Object-based random forest classification for mapping floodplain vegetation structure from nation-wide CIR and LiDAR datasets. In 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014 [8077590] IEEE computer society. https://doi.org/10.1109/WHISPERS.2014.8077590
Kooistra, L. ; Kuilder, E.T. ; Mucher, C.A. / Object-based random forest classification for mapping floodplain vegetation structure from nation-wide CIR and LiDAR datasets. 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014. IEEE computer society, 2017.
@inproceedings{5a29849f473648c4946d2c1c259fc9c3,
title = "Object-based random forest classification for mapping floodplain vegetation structure from nation-wide CIR and LiDAR datasets",
abstract = "Very high resolution aerial images and LiDAR (AHN2) datasets with a national coverage provide opportunities to produce vegetation maps automatically. As such the entire area of the river floodplains in the Netherlands may be mapped with high accuracy and regular updates, capturing the dynamic state of the vegetation. In this study, these fused datasets are used to map the vegetation of 936 ha of the floodplain on the north-side of the river Nederrijn near Wageningen into ten vegetation structure classes. The method follows object-based image analysis principles. Objects are defined in segmentation and subsequently labeled using the ensemble-tree classifier random forest. The mapping scale is controlled by selecting segmentation parameters from quantified discrepancies between reference polygons and segmented objects. Effects on the mapping scale of different reference polygons and different segmentation data is investigated. The results show that it is important to be able to select the right segmentation parameters to control the mapping scale. A discrepancy measure with reference polygons is a suitable method to do this objectively. The use of random forest classification on the objects resulted in an estimated classification accuracy of 86{\%} on the basis of the built-in cross-validation estimate of random forest. Variable importance measures of random forest showed that the AHN2 lidar dataset is a valuable addition to the spectral information contained in the aerial images in the classification.",
keywords = "Object Based Image Analysis (OBIA), reference polygons, segmentation optimization, variable importance, vegetation structure classes",
author = "L. Kooistra and E.T. Kuilder and C.A. Mucher",
year = "2017",
month = "10",
day = "26",
doi = "10.1109/WHISPERS.2014.8077590",
language = "English",
isbn = "9781467390132",
booktitle = "6th Workshop on Hyperspectral Image and Signal Processing",
publisher = "IEEE computer society",

}

Kooistra, L, Kuilder, ET & Mucher, CA 2017, Object-based random forest classification for mapping floodplain vegetation structure from nation-wide CIR and LiDAR datasets. in 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014., 8077590, IEEE computer society, 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014, Lausanne, Switzerland, 24/06/14. https://doi.org/10.1109/WHISPERS.2014.8077590

Object-based random forest classification for mapping floodplain vegetation structure from nation-wide CIR and LiDAR datasets. / Kooistra, L.; Kuilder, E.T.; Mucher, C.A.

6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014. IEEE computer society, 2017. 8077590.

Research output: Chapter in Book/Report/Conference proceedingConference paper

TY - GEN

T1 - Object-based random forest classification for mapping floodplain vegetation structure from nation-wide CIR and LiDAR datasets

AU - Kooistra, L.

AU - Kuilder, E.T.

AU - Mucher, C.A.

PY - 2017/10/26

Y1 - 2017/10/26

N2 - Very high resolution aerial images and LiDAR (AHN2) datasets with a national coverage provide opportunities to produce vegetation maps automatically. As such the entire area of the river floodplains in the Netherlands may be mapped with high accuracy and regular updates, capturing the dynamic state of the vegetation. In this study, these fused datasets are used to map the vegetation of 936 ha of the floodplain on the north-side of the river Nederrijn near Wageningen into ten vegetation structure classes. The method follows object-based image analysis principles. Objects are defined in segmentation and subsequently labeled using the ensemble-tree classifier random forest. The mapping scale is controlled by selecting segmentation parameters from quantified discrepancies between reference polygons and segmented objects. Effects on the mapping scale of different reference polygons and different segmentation data is investigated. The results show that it is important to be able to select the right segmentation parameters to control the mapping scale. A discrepancy measure with reference polygons is a suitable method to do this objectively. The use of random forest classification on the objects resulted in an estimated classification accuracy of 86% on the basis of the built-in cross-validation estimate of random forest. Variable importance measures of random forest showed that the AHN2 lidar dataset is a valuable addition to the spectral information contained in the aerial images in the classification.

AB - Very high resolution aerial images and LiDAR (AHN2) datasets with a national coverage provide opportunities to produce vegetation maps automatically. As such the entire area of the river floodplains in the Netherlands may be mapped with high accuracy and regular updates, capturing the dynamic state of the vegetation. In this study, these fused datasets are used to map the vegetation of 936 ha of the floodplain on the north-side of the river Nederrijn near Wageningen into ten vegetation structure classes. The method follows object-based image analysis principles. Objects are defined in segmentation and subsequently labeled using the ensemble-tree classifier random forest. The mapping scale is controlled by selecting segmentation parameters from quantified discrepancies between reference polygons and segmented objects. Effects on the mapping scale of different reference polygons and different segmentation data is investigated. The results show that it is important to be able to select the right segmentation parameters to control the mapping scale. A discrepancy measure with reference polygons is a suitable method to do this objectively. The use of random forest classification on the objects resulted in an estimated classification accuracy of 86% on the basis of the built-in cross-validation estimate of random forest. Variable importance measures of random forest showed that the AHN2 lidar dataset is a valuable addition to the spectral information contained in the aerial images in the classification.

KW - Object Based Image Analysis (OBIA)

KW - reference polygons

KW - segmentation optimization

KW - variable importance

KW - vegetation structure classes

U2 - 10.1109/WHISPERS.2014.8077590

DO - 10.1109/WHISPERS.2014.8077590

M3 - Conference paper

SN - 9781467390132

BT - 6th Workshop on Hyperspectral Image and Signal Processing

PB - IEEE computer society

ER -

Kooistra L, Kuilder ET, Mucher CA. Object-based random forest classification for mapping floodplain vegetation structure from nation-wide CIR and LiDAR datasets. In 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing, WHISPERS 2014. IEEE computer society. 2017. 8077590 https://doi.org/10.1109/WHISPERS.2014.8077590