Detection by classification of buildings in multispectral satellite imagery

Tomohiro Ishii, Edgar Simo Serra, Satoshi Iizuka, Yoshihiko Mochizuki, Akihiro Sugimoto, Hiroshi Ishikawa, Ryosuke Nakamura

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

12 Citations (Scopus)

Abstract

We present an approach for the detection of buildings in multispectral satellite images. Unlike 3-channel RGB images, satellite imagery contains additional channels corresponding to different wavelengths. Approaches that do not use all channels are unable to fully exploit these images for optimal performance. Furthermore, care must be taken due to the large bias in classes, e.g., most of the Earth is covered in water and thus it will be dominant in the images. Our approach consists of training a Convolutional Neural Network (CNN) from scratch to classify multispectral image patches taken by satellites as whether or not they belong to a class of buildings. We then adapt the classification network to detection by converting the fully-connected layers of the network to convolutional layers, which allows the network to process images of any resolution. The dataset bias is compensated by subsampling negatives and tuning the detection threshold for optimal performance. We have constructed a new dataset using images from the Landsat 8 satellite for detecting solar power plants and show our approach is able to significantly outperform the state-of-the-art. Furthermore, we provide an indepth evaluation of the seven different spectral bands provided by the satellite images and show it is critical to combine them to obtain good results.

Original languageEnglish
Title of host publication2016 23rd International Conference on Pattern Recognition, ICPR 2016
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages3344-3349
Number of pages6
ISBN (Electronic)9781509048472
DOIs
Publication statusPublished - 2017 Apr 13
Event23rd International Conference on Pattern Recognition, ICPR 2016 - Cancun, Mexico
Duration: 2016 Dec 42016 Dec 8

Other

Other23rd International Conference on Pattern Recognition, ICPR 2016
CountryMexico
CityCancun
Period16/12/416/12/8

Fingerprint

Satellite imagery
Satellites
Solar power plants
Tuning
Earth (planet)
Neural networks
Wavelength
Water

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition

Cite this

Ishii, T., Simo Serra, E., Iizuka, S., Mochizuki, Y., Sugimoto, A., Ishikawa, H., & Nakamura, R. (2017). Detection by classification of buildings in multispectral satellite imagery. In 2016 23rd International Conference on Pattern Recognition, ICPR 2016 (pp. 3344-3349). [7900150] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICPR.2016.7900150

Detection by classification of buildings in multispectral satellite imagery. / Ishii, Tomohiro; Simo Serra, Edgar; Iizuka, Satoshi; Mochizuki, Yoshihiko; Sugimoto, Akihiro; Ishikawa, Hiroshi; Nakamura, Ryosuke.

2016 23rd International Conference on Pattern Recognition, ICPR 2016. Institute of Electrical and Electronics Engineers Inc., 2017. p. 3344-3349 7900150.

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

Ishii, T, Simo Serra, E, Iizuka, S, Mochizuki, Y, Sugimoto, A, Ishikawa, H & Nakamura, R 2017, Detection by classification of buildings in multispectral satellite imagery. in 2016 23rd International Conference on Pattern Recognition, ICPR 2016., 7900150, Institute of Electrical and Electronics Engineers Inc., pp. 3344-3349, 23rd International Conference on Pattern Recognition, ICPR 2016, Cancun, Mexico, 16/12/4. https://doi.org/10.1109/ICPR.2016.7900150
Ishii T, Simo Serra E, Iizuka S, Mochizuki Y, Sugimoto A, Ishikawa H et al. Detection by classification of buildings in multispectral satellite imagery. In 2016 23rd International Conference on Pattern Recognition, ICPR 2016. Institute of Electrical and Electronics Engineers Inc. 2017. p. 3344-3349. 7900150 https://doi.org/10.1109/ICPR.2016.7900150
Ishii, Tomohiro ; Simo Serra, Edgar ; Iizuka, Satoshi ; Mochizuki, Yoshihiko ; Sugimoto, Akihiro ; Ishikawa, Hiroshi ; Nakamura, Ryosuke. / Detection by classification of buildings in multispectral satellite imagery. 2016 23rd International Conference on Pattern Recognition, ICPR 2016. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 3344-3349
@inproceedings{b102b88d8668410990a5ced3f5bbed25,
title = "Detection by classification of buildings in multispectral satellite imagery",
abstract = "We present an approach for the detection of buildings in multispectral satellite images. Unlike 3-channel RGB images, satellite imagery contains additional channels corresponding to different wavelengths. Approaches that do not use all channels are unable to fully exploit these images for optimal performance. Furthermore, care must be taken due to the large bias in classes, e.g., most of the Earth is covered in water and thus it will be dominant in the images. Our approach consists of training a Convolutional Neural Network (CNN) from scratch to classify multispectral image patches taken by satellites as whether or not they belong to a class of buildings. We then adapt the classification network to detection by converting the fully-connected layers of the network to convolutional layers, which allows the network to process images of any resolution. The dataset bias is compensated by subsampling negatives and tuning the detection threshold for optimal performance. We have constructed a new dataset using images from the Landsat 8 satellite for detecting solar power plants and show our approach is able to significantly outperform the state-of-the-art. Furthermore, we provide an indepth evaluation of the seven different spectral bands provided by the satellite images and show it is critical to combine them to obtain good results.",
author = "Tomohiro Ishii and {Simo Serra}, Edgar and Satoshi Iizuka and Yoshihiko Mochizuki and Akihiro Sugimoto and Hiroshi Ishikawa and Ryosuke Nakamura",
year = "2017",
month = "4",
day = "13",
doi = "10.1109/ICPR.2016.7900150",
language = "English",
pages = "3344--3349",
booktitle = "2016 23rd International Conference on Pattern Recognition, ICPR 2016",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
address = "United States",

}

TY - GEN

T1 - Detection by classification of buildings in multispectral satellite imagery

AU - Ishii, Tomohiro

AU - Simo Serra, Edgar

AU - Iizuka, Satoshi

AU - Mochizuki, Yoshihiko

AU - Sugimoto, Akihiro

AU - Ishikawa, Hiroshi

AU - Nakamura, Ryosuke

PY - 2017/4/13

Y1 - 2017/4/13

N2 - We present an approach for the detection of buildings in multispectral satellite images. Unlike 3-channel RGB images, satellite imagery contains additional channels corresponding to different wavelengths. Approaches that do not use all channels are unable to fully exploit these images for optimal performance. Furthermore, care must be taken due to the large bias in classes, e.g., most of the Earth is covered in water and thus it will be dominant in the images. Our approach consists of training a Convolutional Neural Network (CNN) from scratch to classify multispectral image patches taken by satellites as whether or not they belong to a class of buildings. We then adapt the classification network to detection by converting the fully-connected layers of the network to convolutional layers, which allows the network to process images of any resolution. The dataset bias is compensated by subsampling negatives and tuning the detection threshold for optimal performance. We have constructed a new dataset using images from the Landsat 8 satellite for detecting solar power plants and show our approach is able to significantly outperform the state-of-the-art. Furthermore, we provide an indepth evaluation of the seven different spectral bands provided by the satellite images and show it is critical to combine them to obtain good results.

AB - We present an approach for the detection of buildings in multispectral satellite images. Unlike 3-channel RGB images, satellite imagery contains additional channels corresponding to different wavelengths. Approaches that do not use all channels are unable to fully exploit these images for optimal performance. Furthermore, care must be taken due to the large bias in classes, e.g., most of the Earth is covered in water and thus it will be dominant in the images. Our approach consists of training a Convolutional Neural Network (CNN) from scratch to classify multispectral image patches taken by satellites as whether or not they belong to a class of buildings. We then adapt the classification network to detection by converting the fully-connected layers of the network to convolutional layers, which allows the network to process images of any resolution. The dataset bias is compensated by subsampling negatives and tuning the detection threshold for optimal performance. We have constructed a new dataset using images from the Landsat 8 satellite for detecting solar power plants and show our approach is able to significantly outperform the state-of-the-art. Furthermore, we provide an indepth evaluation of the seven different spectral bands provided by the satellite images and show it is critical to combine them to obtain good results.

UR - http://www.scopus.com/inward/record.url?scp=85019106511&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85019106511&partnerID=8YFLogxK

U2 - 10.1109/ICPR.2016.7900150

DO - 10.1109/ICPR.2016.7900150

M3 - Conference contribution

SP - 3344

EP - 3349

BT - 2016 23rd International Conference on Pattern Recognition, ICPR 2016

PB - Institute of Electrical and Electronics Engineers Inc.

ER -