Separation of coniferous species in boreal forest using spectral and contextual features from ikonos imagery

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

Abstract

Trees were located and classified to pine and spruce classes using features computed from Ikonos multispectral and panchromatic channels at study site in Eastern Finland. Spectral signatures were sampled from the extracted tree locations, and a set of contextual features was computed in the neighborhood around each located tree from the Ikonos panchromatic channel. Circular masks of five different sizes were used. The contextual features included higher order statistical features (skewness, kurtosis), and additional features obtained by fitting either two Gaussian distributions or a Weibull distribution to the intensity histogram. The contextual features aimed at capturing differences in the distribution of intensities of pine and spruce crowns. Pure (100%) pine and spruce plots with medium stem volume (100 - 200 m3/ha), and with a minimum distance of 15 m from stand borders, were used in the study. The training data contained 5 plots of pine and 5 plots of spruce, from which 196 trees were located (96 pine, 100 spruce). The separate validation data set consisted of 6 plots (3 plots both pine and spruce) containing 116 trees (54 pine, 62 spruce). Stepwise linear discriminant analysis was used to select the best separating features and for classification. From the multispectral channels, the best separating feature was the blue channel. From the contextual features the best separating features were the Weibull shape parameter, the ratio of sample mean and median, kurtosis, and skewness, the set of best features being slightly different for different sampling radii. For the validation data set, the percentage of correctly classified trees was 87.9% when using spectral channels only, and increased from 81.9% to 98.3% along with increasing sampling radius using only the contextual features. The classification accuracy reached 99.1% when both spectral and contextual features were used.

Original languageEnglish
Title of host publication2006 IEEE International Geoscience and Remote Sensing Symposium, IGARSS
Pages2141-2144
Number of pages4
Volume6
DOIs
Publication statusPublished - 1 Dec 2006
MoE publication typeA4 Article in a conference publication
EventIEEE International Geoscience and Remote Sensing Symposium, IGARSS 2006 - Denver, CO, United States
Duration: 31 Jul 20064 Aug 2006

Conference

ConferenceIEEE International Geoscience and Remote Sensing Symposium, IGARSS 2006
Abbreviated titleIGARSS 2006
CountryUnited States
CityDenver, CO
Period31/07/064/08/06

Fingerprint

boreal forest
imagery
Sampling
Weibull distribution
Gaussian distribution
Discriminant analysis
Masks
skewness
sampling
histogram
discriminant analysis
stem
distribution

Keywords

  • Forest
  • High resolution
  • Ikonos
  • Image analysis
  • Remote sensing
  • Species classification
  • Texture
  • Tree species

Cite this

Astola, H., Sirro, L., Häme, T., Molinier, M., & Ahola, J. (2006). Separation of coniferous species in boreal forest using spectral and contextual features from ikonos imagery. In 2006 IEEE International Geoscience and Remote Sensing Symposium, IGARSS (Vol. 6, pp. 2141-2144). [4241701] https://doi.org/10.1109/IGARSS.2006.554
Astola, Heikki ; Sirro, Laura ; Häme, Tuomas ; Molinier, Matthieu ; Ahola, Jussi. / Separation of coniferous species in boreal forest using spectral and contextual features from ikonos imagery. 2006 IEEE International Geoscience and Remote Sensing Symposium, IGARSS. Vol. 6 2006. pp. 2141-2144
@inproceedings{e2d5e8860fc84a538dbfbdfe6c74e12a,
title = "Separation of coniferous species in boreal forest using spectral and contextual features from ikonos imagery",
abstract = "Trees were located and classified to pine and spruce classes using features computed from Ikonos multispectral and panchromatic channels at study site in Eastern Finland. Spectral signatures were sampled from the extracted tree locations, and a set of contextual features was computed in the neighborhood around each located tree from the Ikonos panchromatic channel. Circular masks of five different sizes were used. The contextual features included higher order statistical features (skewness, kurtosis), and additional features obtained by fitting either two Gaussian distributions or a Weibull distribution to the intensity histogram. The contextual features aimed at capturing differences in the distribution of intensities of pine and spruce crowns. Pure (100{\%}) pine and spruce plots with medium stem volume (100 - 200 m3/ha), and with a minimum distance of 15 m from stand borders, were used in the study. The training data contained 5 plots of pine and 5 plots of spruce, from which 196 trees were located (96 pine, 100 spruce). The separate validation data set consisted of 6 plots (3 plots both pine and spruce) containing 116 trees (54 pine, 62 spruce). Stepwise linear discriminant analysis was used to select the best separating features and for classification. From the multispectral channels, the best separating feature was the blue channel. From the contextual features the best separating features were the Weibull shape parameter, the ratio of sample mean and median, kurtosis, and skewness, the set of best features being slightly different for different sampling radii. For the validation data set, the percentage of correctly classified trees was 87.9{\%} when using spectral channels only, and increased from 81.9{\%} to 98.3{\%} along with increasing sampling radius using only the contextual features. The classification accuracy reached 99.1{\%} when both spectral and contextual features were used.",
keywords = "Forest, High resolution, Ikonos, Image analysis, Remote sensing, Species classification, Texture, Tree species",
author = "Heikki Astola and Laura Sirro and Tuomas H{\"a}me and Matthieu Molinier and Jussi Ahola",
year = "2006",
month = "12",
day = "1",
doi = "10.1109/IGARSS.2006.554",
language = "English",
isbn = "0780395107",
volume = "6",
pages = "2141--2144",
booktitle = "2006 IEEE International Geoscience and Remote Sensing Symposium, IGARSS",

}

Astola, H, Sirro, L, Häme, T, Molinier, M & Ahola, J 2006, Separation of coniferous species in boreal forest using spectral and contextual features from ikonos imagery. in 2006 IEEE International Geoscience and Remote Sensing Symposium, IGARSS. vol. 6, 4241701, pp. 2141-2144, IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2006, Denver, CO, United States, 31/07/06. https://doi.org/10.1109/IGARSS.2006.554

Separation of coniferous species in boreal forest using spectral and contextual features from ikonos imagery. / Astola, Heikki; Sirro, Laura; Häme, Tuomas; Molinier, Matthieu; Ahola, Jussi.

2006 IEEE International Geoscience and Remote Sensing Symposium, IGARSS. Vol. 6 2006. p. 2141-2144 4241701.

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

TY - GEN

T1 - Separation of coniferous species in boreal forest using spectral and contextual features from ikonos imagery

AU - Astola, Heikki

AU - Sirro, Laura

AU - Häme, Tuomas

AU - Molinier, Matthieu

AU - Ahola, Jussi

PY - 2006/12/1

Y1 - 2006/12/1

N2 - Trees were located and classified to pine and spruce classes using features computed from Ikonos multispectral and panchromatic channels at study site in Eastern Finland. Spectral signatures were sampled from the extracted tree locations, and a set of contextual features was computed in the neighborhood around each located tree from the Ikonos panchromatic channel. Circular masks of five different sizes were used. The contextual features included higher order statistical features (skewness, kurtosis), and additional features obtained by fitting either two Gaussian distributions or a Weibull distribution to the intensity histogram. The contextual features aimed at capturing differences in the distribution of intensities of pine and spruce crowns. Pure (100%) pine and spruce plots with medium stem volume (100 - 200 m3/ha), and with a minimum distance of 15 m from stand borders, were used in the study. The training data contained 5 plots of pine and 5 plots of spruce, from which 196 trees were located (96 pine, 100 spruce). The separate validation data set consisted of 6 plots (3 plots both pine and spruce) containing 116 trees (54 pine, 62 spruce). Stepwise linear discriminant analysis was used to select the best separating features and for classification. From the multispectral channels, the best separating feature was the blue channel. From the contextual features the best separating features were the Weibull shape parameter, the ratio of sample mean and median, kurtosis, and skewness, the set of best features being slightly different for different sampling radii. For the validation data set, the percentage of correctly classified trees was 87.9% when using spectral channels only, and increased from 81.9% to 98.3% along with increasing sampling radius using only the contextual features. The classification accuracy reached 99.1% when both spectral and contextual features were used.

AB - Trees were located and classified to pine and spruce classes using features computed from Ikonos multispectral and panchromatic channels at study site in Eastern Finland. Spectral signatures were sampled from the extracted tree locations, and a set of contextual features was computed in the neighborhood around each located tree from the Ikonos panchromatic channel. Circular masks of five different sizes were used. The contextual features included higher order statistical features (skewness, kurtosis), and additional features obtained by fitting either two Gaussian distributions or a Weibull distribution to the intensity histogram. The contextual features aimed at capturing differences in the distribution of intensities of pine and spruce crowns. Pure (100%) pine and spruce plots with medium stem volume (100 - 200 m3/ha), and with a minimum distance of 15 m from stand borders, were used in the study. The training data contained 5 plots of pine and 5 plots of spruce, from which 196 trees were located (96 pine, 100 spruce). The separate validation data set consisted of 6 plots (3 plots both pine and spruce) containing 116 trees (54 pine, 62 spruce). Stepwise linear discriminant analysis was used to select the best separating features and for classification. From the multispectral channels, the best separating feature was the blue channel. From the contextual features the best separating features were the Weibull shape parameter, the ratio of sample mean and median, kurtosis, and skewness, the set of best features being slightly different for different sampling radii. For the validation data set, the percentage of correctly classified trees was 87.9% when using spectral channels only, and increased from 81.9% to 98.3% along with increasing sampling radius using only the contextual features. The classification accuracy reached 99.1% when both spectral and contextual features were used.

KW - Forest

KW - High resolution

KW - Ikonos

KW - Image analysis

KW - Remote sensing

KW - Species classification

KW - Texture

KW - Tree species

UR - http://www.scopus.com/inward/record.url?scp=33947433596&partnerID=8YFLogxK

U2 - 10.1109/IGARSS.2006.554

DO - 10.1109/IGARSS.2006.554

M3 - Conference article in proceedings

AN - SCOPUS:34948873020

SN - 0780395107

SN - 9780780395107

VL - 6

SP - 2141

EP - 2144

BT - 2006 IEEE International Geoscience and Remote Sensing Symposium, IGARSS

ER -

Astola H, Sirro L, Häme T, Molinier M, Ahola J. Separation of coniferous species in boreal forest using spectral and contextual features from ikonos imagery. In 2006 IEEE International Geoscience and Remote Sensing Symposium, IGARSS. Vol. 6. 2006. p. 2141-2144. 4241701 https://doi.org/10.1109/IGARSS.2006.554