Semantic segmentation and difference extraction via time series aerial video camera and its application

S. N K Amit, S. Saito, S. Sasaki, Yasushi Kiyoki, Yoshimitsu Aoki

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

1 Citation (Scopus)

Abstract

Google earth with high-resolution imagery basically takes months to process new images before online updates. It is a time consuming and slow process especially for post-disaster application. The objective of this research is to develop a fast and effective method of updating maps by detecting local differences occurred over different time series; where only region with differences will be updated. In our system, aerial images from Massachusetts's road and building open datasets, Saitama district datasets are used as input images. Semantic segmentation is then applied to input images. Semantic segmentation is a pixel-wise classification of images by implementing deep neural network technique. Deep neural network technique is implemented due to being not only efficient in learning highly discriminative image features such as road, buildings etc., but also partially robust to incomplete and poorly registered target maps. Then, aerial images which contain semantic information are stored as database in 5D world map is set as ground truth images. This system is developed to visualise multimedia data in 5 dimensions; 3 dimensions as spatial dimensions, 1 dimension as temporal dimension, and 1 dimension as degenerated dimensions of semantic and colour combination dimension. Next, ground truth images chosen from database in 5D world map and a new aerial image with same spatial information but different time series are compared via difference extraction method. The map will only update where local changes had occurred. Hence, map updating will be cheaper, faster and more effective especially post-disaster application, by leaving unchanged region and only update changed region.

Original languageEnglish
Title of host publicationInternational Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives
PublisherInternational Society for Photogrammetry and Remote Sensing
Pages1119-1122
Number of pages4
Volume40
Edition7W3
DOIs
Publication statusPublished - 2015 Apr 28
Event2015 36th International Symposium on Remote Sensing of Environment - Berlin, Germany
Duration: 2015 May 112015 May 15

Other

Other2015 36th International Symposium on Remote Sensing of Environment
CountryGermany
CityBerlin
Period15/5/1115/5/15

Fingerprint

Video cameras
segmentation
time series
Time series
video
Semantics
semantics
Antennas
Disasters
neural network
disaster
road
Pixels
Earth (planet)
Color
multimedia
search engine
extraction method
building
pixel

Keywords

  • Aerial images
  • Convolution neural networks
  • Difference extraction
  • Semantic segmentation

ASJC Scopus subject areas

  • Information Systems
  • Geography, Planning and Development

Cite this

Amit, S. N. K., Saito, S., Sasaki, S., Kiyoki, Y., & Aoki, Y. (2015). Semantic segmentation and difference extraction via time series aerial video camera and its application. In International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives (7W3 ed., Vol. 40, pp. 1119-1122). International Society for Photogrammetry and Remote Sensing. https://doi.org/10.5194/isprsarchives-XL-7-W3-1119-2015

Semantic segmentation and difference extraction via time series aerial video camera and its application. / Amit, S. N K; Saito, S.; Sasaki, S.; Kiyoki, Yasushi; Aoki, Yoshimitsu.

International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives. Vol. 40 7W3. ed. International Society for Photogrammetry and Remote Sensing, 2015. p. 1119-1122.

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

Amit, SNK, Saito, S, Sasaki, S, Kiyoki, Y & Aoki, Y 2015, Semantic segmentation and difference extraction via time series aerial video camera and its application. in International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives. 7W3 edn, vol. 40, International Society for Photogrammetry and Remote Sensing, pp. 1119-1122, 2015 36th International Symposium on Remote Sensing of Environment, Berlin, Germany, 15/5/11. https://doi.org/10.5194/isprsarchives-XL-7-W3-1119-2015
Amit SNK, Saito S, Sasaki S, Kiyoki Y, Aoki Y. Semantic segmentation and difference extraction via time series aerial video camera and its application. In International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives. 7W3 ed. Vol. 40. International Society for Photogrammetry and Remote Sensing. 2015. p. 1119-1122 https://doi.org/10.5194/isprsarchives-XL-7-W3-1119-2015
Amit, S. N K ; Saito, S. ; Sasaki, S. ; Kiyoki, Yasushi ; Aoki, Yoshimitsu. / Semantic segmentation and difference extraction via time series aerial video camera and its application. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives. Vol. 40 7W3. ed. International Society for Photogrammetry and Remote Sensing, 2015. pp. 1119-1122
@inproceedings{5bfef2e241e146288427f1cf1e094521,
title = "Semantic segmentation and difference extraction via time series aerial video camera and its application",
abstract = "Google earth with high-resolution imagery basically takes months to process new images before online updates. It is a time consuming and slow process especially for post-disaster application. The objective of this research is to develop a fast and effective method of updating maps by detecting local differences occurred over different time series; where only region with differences will be updated. In our system, aerial images from Massachusetts's road and building open datasets, Saitama district datasets are used as input images. Semantic segmentation is then applied to input images. Semantic segmentation is a pixel-wise classification of images by implementing deep neural network technique. Deep neural network technique is implemented due to being not only efficient in learning highly discriminative image features such as road, buildings etc., but also partially robust to incomplete and poorly registered target maps. Then, aerial images which contain semantic information are stored as database in 5D world map is set as ground truth images. This system is developed to visualise multimedia data in 5 dimensions; 3 dimensions as spatial dimensions, 1 dimension as temporal dimension, and 1 dimension as degenerated dimensions of semantic and colour combination dimension. Next, ground truth images chosen from database in 5D world map and a new aerial image with same spatial information but different time series are compared via difference extraction method. The map will only update where local changes had occurred. Hence, map updating will be cheaper, faster and more effective especially post-disaster application, by leaving unchanged region and only update changed region.",
keywords = "Aerial images, Convolution neural networks, Difference extraction, Semantic segmentation",
author = "Amit, {S. N K} and S. Saito and S. Sasaki and Yasushi Kiyoki and Yoshimitsu Aoki",
year = "2015",
month = "4",
day = "28",
doi = "10.5194/isprsarchives-XL-7-W3-1119-2015",
language = "English",
volume = "40",
pages = "1119--1122",
booktitle = "International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives",
publisher = "International Society for Photogrammetry and Remote Sensing",
edition = "7W3",

}

TY - GEN

T1 - Semantic segmentation and difference extraction via time series aerial video camera and its application

AU - Amit, S. N K

AU - Saito, S.

AU - Sasaki, S.

AU - Kiyoki, Yasushi

AU - Aoki, Yoshimitsu

PY - 2015/4/28

Y1 - 2015/4/28

N2 - Google earth with high-resolution imagery basically takes months to process new images before online updates. It is a time consuming and slow process especially for post-disaster application. The objective of this research is to develop a fast and effective method of updating maps by detecting local differences occurred over different time series; where only region with differences will be updated. In our system, aerial images from Massachusetts's road and building open datasets, Saitama district datasets are used as input images. Semantic segmentation is then applied to input images. Semantic segmentation is a pixel-wise classification of images by implementing deep neural network technique. Deep neural network technique is implemented due to being not only efficient in learning highly discriminative image features such as road, buildings etc., but also partially robust to incomplete and poorly registered target maps. Then, aerial images which contain semantic information are stored as database in 5D world map is set as ground truth images. This system is developed to visualise multimedia data in 5 dimensions; 3 dimensions as spatial dimensions, 1 dimension as temporal dimension, and 1 dimension as degenerated dimensions of semantic and colour combination dimension. Next, ground truth images chosen from database in 5D world map and a new aerial image with same spatial information but different time series are compared via difference extraction method. The map will only update where local changes had occurred. Hence, map updating will be cheaper, faster and more effective especially post-disaster application, by leaving unchanged region and only update changed region.

AB - Google earth with high-resolution imagery basically takes months to process new images before online updates. It is a time consuming and slow process especially for post-disaster application. The objective of this research is to develop a fast and effective method of updating maps by detecting local differences occurred over different time series; where only region with differences will be updated. In our system, aerial images from Massachusetts's road and building open datasets, Saitama district datasets are used as input images. Semantic segmentation is then applied to input images. Semantic segmentation is a pixel-wise classification of images by implementing deep neural network technique. Deep neural network technique is implemented due to being not only efficient in learning highly discriminative image features such as road, buildings etc., but also partially robust to incomplete and poorly registered target maps. Then, aerial images which contain semantic information are stored as database in 5D world map is set as ground truth images. This system is developed to visualise multimedia data in 5 dimensions; 3 dimensions as spatial dimensions, 1 dimension as temporal dimension, and 1 dimension as degenerated dimensions of semantic and colour combination dimension. Next, ground truth images chosen from database in 5D world map and a new aerial image with same spatial information but different time series are compared via difference extraction method. The map will only update where local changes had occurred. Hence, map updating will be cheaper, faster and more effective especially post-disaster application, by leaving unchanged region and only update changed region.

KW - Aerial images

KW - Convolution neural networks

KW - Difference extraction

KW - Semantic segmentation

UR - http://www.scopus.com/inward/record.url?scp=84930405236&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84930405236&partnerID=8YFLogxK

U2 - 10.5194/isprsarchives-XL-7-W3-1119-2015

DO - 10.5194/isprsarchives-XL-7-W3-1119-2015

M3 - Conference contribution

AN - SCOPUS:84930405236

VL - 40

SP - 1119

EP - 1122

BT - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives

PB - International Society for Photogrammetry and Remote Sensing

ER -