Study of automatic image matching and registration of scanned historical aerial photographs

Hou Ren Chen, Yi-Hsing Tseng

Research output: Contribution to conferencePaper

Abstract

Historical aerial images provide good evidences for the analysis of temporal environmental changes. The Research Center for Humanities and Social Sciences (RCHSS), Taiwan Academia Sinica, has collected and scanned numerous historical maps and aerial images of Taiwan. Some images have been processed manually for geo-referencing, but most of them have not been registered since they have no GPS or IMU data for orientation assisting in the past. After digital scanning, those historical aerial images can be processed by methods of computer vision. For example, extracting features of different aerial images, or using dense image matching for automatic reconstruction of buildings. The concept of image matching is to find similar features in different images. Two images are considered to be overlapped if they have enough same features, and the orientation parameters of different images can be derived. With the relative orientation, it's possible to mosaic or locate images. Many recent photogrammetric software use image matching to stitch photographs into a panoramic image or build 3D model. In our study, the software used to stitch images are PTGui and Agisoft PhotoScan. Both of them can align images automatically. PTGui has an approachable user interface, and it can extract image features, match image features and align images effectively. The second software, Agisoft PhotoScan, has something different in the processing of final stitched image. After images alignment, it builds dense clouds and then builds irregular mesh for generating 3D model. Finally, we can acquire orthoimage which is stitched from original input images. The purpose of this research are assessments of stitched historical aerial images obtained from two software, accuracy of image locating, and their efficiency of automation. And we try to geo-reference the stitched images by giving control points. So historical environmental changes can be investigated with those geo-referenced temporal spatial data.

Original languageEnglish
Publication statusPublished - 2015 Jan 1
Event36th Asian Conference on Remote Sensing: Fostering Resilient Growth in Asia, ACRS 2015 - Quezon City, Metro Manila, Philippines
Duration: 2015 Oct 242015 Oct 28

Other

Other36th Asian Conference on Remote Sensing: Fostering Resilient Growth in Asia, ACRS 2015
CountryPhilippines
CityQuezon City, Metro Manila
Period15-10-2415-10-28

Fingerprint

Image matching
Image registration
Antennas
Social sciences
Computer vision
User interfaces
Global positioning system
Automation
Scanning
Processing

All Science Journal Classification (ASJC) codes

  • Computer Networks and Communications

Cite this

Chen, H. R., & Tseng, Y-H. (2015). Study of automatic image matching and registration of scanned historical aerial photographs. Paper presented at 36th Asian Conference on Remote Sensing: Fostering Resilient Growth in Asia, ACRS 2015, Quezon City, Metro Manila, Philippines.
Chen, Hou Ren ; Tseng, Yi-Hsing. / Study of automatic image matching and registration of scanned historical aerial photographs. Paper presented at 36th Asian Conference on Remote Sensing: Fostering Resilient Growth in Asia, ACRS 2015, Quezon City, Metro Manila, Philippines.
@conference{b042cd4fe4a844b7ba6e33a1b2a0343a,
title = "Study of automatic image matching and registration of scanned historical aerial photographs",
abstract = "Historical aerial images provide good evidences for the analysis of temporal environmental changes. The Research Center for Humanities and Social Sciences (RCHSS), Taiwan Academia Sinica, has collected and scanned numerous historical maps and aerial images of Taiwan. Some images have been processed manually for geo-referencing, but most of them have not been registered since they have no GPS or IMU data for orientation assisting in the past. After digital scanning, those historical aerial images can be processed by methods of computer vision. For example, extracting features of different aerial images, or using dense image matching for automatic reconstruction of buildings. The concept of image matching is to find similar features in different images. Two images are considered to be overlapped if they have enough same features, and the orientation parameters of different images can be derived. With the relative orientation, it's possible to mosaic or locate images. Many recent photogrammetric software use image matching to stitch photographs into a panoramic image or build 3D model. In our study, the software used to stitch images are PTGui and Agisoft PhotoScan. Both of them can align images automatically. PTGui has an approachable user interface, and it can extract image features, match image features and align images effectively. The second software, Agisoft PhotoScan, has something different in the processing of final stitched image. After images alignment, it builds dense clouds and then builds irregular mesh for generating 3D model. Finally, we can acquire orthoimage which is stitched from original input images. The purpose of this research are assessments of stitched historical aerial images obtained from two software, accuracy of image locating, and their efficiency of automation. And we try to geo-reference the stitched images by giving control points. So historical environmental changes can be investigated with those geo-referenced temporal spatial data.",
author = "Chen, {Hou Ren} and Yi-Hsing Tseng",
year = "2015",
month = "1",
day = "1",
language = "English",
note = "36th Asian Conference on Remote Sensing: Fostering Resilient Growth in Asia, ACRS 2015 ; Conference date: 24-10-2015 Through 28-10-2015",

}

Chen, HR & Tseng, Y-H 2015, 'Study of automatic image matching and registration of scanned historical aerial photographs' Paper presented at 36th Asian Conference on Remote Sensing: Fostering Resilient Growth in Asia, ACRS 2015, Quezon City, Metro Manila, Philippines, 15-10-24 - 15-10-28, .

Study of automatic image matching and registration of scanned historical aerial photographs. / Chen, Hou Ren; Tseng, Yi-Hsing.

2015. Paper presented at 36th Asian Conference on Remote Sensing: Fostering Resilient Growth in Asia, ACRS 2015, Quezon City, Metro Manila, Philippines.

Research output: Contribution to conferencePaper

TY - CONF

T1 - Study of automatic image matching and registration of scanned historical aerial photographs

AU - Chen, Hou Ren

AU - Tseng, Yi-Hsing

PY - 2015/1/1

Y1 - 2015/1/1

N2 - Historical aerial images provide good evidences for the analysis of temporal environmental changes. The Research Center for Humanities and Social Sciences (RCHSS), Taiwan Academia Sinica, has collected and scanned numerous historical maps and aerial images of Taiwan. Some images have been processed manually for geo-referencing, but most of them have not been registered since they have no GPS or IMU data for orientation assisting in the past. After digital scanning, those historical aerial images can be processed by methods of computer vision. For example, extracting features of different aerial images, or using dense image matching for automatic reconstruction of buildings. The concept of image matching is to find similar features in different images. Two images are considered to be overlapped if they have enough same features, and the orientation parameters of different images can be derived. With the relative orientation, it's possible to mosaic or locate images. Many recent photogrammetric software use image matching to stitch photographs into a panoramic image or build 3D model. In our study, the software used to stitch images are PTGui and Agisoft PhotoScan. Both of them can align images automatically. PTGui has an approachable user interface, and it can extract image features, match image features and align images effectively. The second software, Agisoft PhotoScan, has something different in the processing of final stitched image. After images alignment, it builds dense clouds and then builds irregular mesh for generating 3D model. Finally, we can acquire orthoimage which is stitched from original input images. The purpose of this research are assessments of stitched historical aerial images obtained from two software, accuracy of image locating, and their efficiency of automation. And we try to geo-reference the stitched images by giving control points. So historical environmental changes can be investigated with those geo-referenced temporal spatial data.

AB - Historical aerial images provide good evidences for the analysis of temporal environmental changes. The Research Center for Humanities and Social Sciences (RCHSS), Taiwan Academia Sinica, has collected and scanned numerous historical maps and aerial images of Taiwan. Some images have been processed manually for geo-referencing, but most of them have not been registered since they have no GPS or IMU data for orientation assisting in the past. After digital scanning, those historical aerial images can be processed by methods of computer vision. For example, extracting features of different aerial images, or using dense image matching for automatic reconstruction of buildings. The concept of image matching is to find similar features in different images. Two images are considered to be overlapped if they have enough same features, and the orientation parameters of different images can be derived. With the relative orientation, it's possible to mosaic or locate images. Many recent photogrammetric software use image matching to stitch photographs into a panoramic image or build 3D model. In our study, the software used to stitch images are PTGui and Agisoft PhotoScan. Both of them can align images automatically. PTGui has an approachable user interface, and it can extract image features, match image features and align images effectively. The second software, Agisoft PhotoScan, has something different in the processing of final stitched image. After images alignment, it builds dense clouds and then builds irregular mesh for generating 3D model. Finally, we can acquire orthoimage which is stitched from original input images. The purpose of this research are assessments of stitched historical aerial images obtained from two software, accuracy of image locating, and their efficiency of automation. And we try to geo-reference the stitched images by giving control points. So historical environmental changes can be investigated with those geo-referenced temporal spatial data.

UR - http://www.scopus.com/inward/record.url?scp=84964068123&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84964068123&partnerID=8YFLogxK

M3 - Paper

ER -

Chen HR, Tseng Y-H. Study of automatic image matching and registration of scanned historical aerial photographs. 2015. Paper presented at 36th Asian Conference on Remote Sensing: Fostering Resilient Growth in Asia, ACRS 2015, Quezon City, Metro Manila, Philippines.