23
E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT Automated processing of high resolution airborne images for earthquake damage assessment Ewelina RUPNIK 3D Optical Metrology (3DOM) Bruno Kessler Foundation (FBK) Trento, Italy Email: [email protected] http://3dom.fbk.eu

Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT

Automated processing of high resolution airborne images for earthquake

damage assessment

Ewelina RUPNIK

3D Optical Metrology (3DOM)

Bruno Kessler Foundation (FBK)

Trento, Italy

Email: [email protected]

http://3dom.fbk.eu

Page 2: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 2

OUTLINE

Introduction

Airborne imaging platforms

Photogrammetric processing

Earthquake damage assessment

Methods

First results

Conclusion and future work

Page 3: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 3

INTRODUCTION

Optical imagery for earthquake damage assessment

Different temporal and spatial resolutions (airborne, spaceborne)

Pre- and/or post-event data

Use of ancillary data, fusion (LiDAR, maps, GIS)

2D and/or 3D approaches

Nadir and/or oblique imagery

Page 4: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 4

INTRODUCTION

Airborne imaging platforms for earthquake damage assessment

Traditional aerial flights e.g. [Li et al., 2014; Gerke and Kerle, 2011;]

San Francisco Earthquake, 1906 – images from kite-borne camera

Wenchuan earthquake 2010, China – nadir views

Haiti earthquake 2010, Dominican Rep. – nadir and oblique views

UAV e.g. [Chou et al., 2010; Huber 2010; Adams and Friedland, 2011]

Hurricane Katrina 2006, Mississipi Gulf, USA

L’Aquila earthquake 2010, Italy

Haiti earthquake 2010, Dominican Rep.

Japan earthquakes 2011

Page 5: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT

PHOTO-INTERPRETATION apt for non-experts

COMPLETENESS 2.5 versus 3D

COST & ACESSIBILITY systematic acquisition, growing market:

Pictometry, MIDAS, DigiCAM, UltraCAM Osprey,

Leica RCD30

Airborne imaging platforms for earthquake damage assessment

Traditional aerial flights e.g. [Li et al., 2014; Gerke and Kerle, 2011;]

San Francisco Earthquake, 1906 – images from kite-borne camera

Wenchuan earthquake 2010, China – nadir views

Haiti earthquake 2010, Dominican Rep. – nadir and oblique views

UAV e.g. [Chou et al., 2010; Huber 2010; Adams and Friedland, 2011]

Hurricane Katrina 2006, Mississipi Gulf, USA

L’Aquila earthquake 2010, Italy

Haiti earthquake 2010, Dominican Rep.

Japan earthquakes 2011

5

INTRODUCTION

Page 6: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 6

OBJECTIVE

Earthquake–induced building damage

area, grade of damage, type of damage

European Macroseismic Scale 1998 (EMS98)

Grade 1 Grade 2 Grade 3 Grade 4 Grade 5

Page 7: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 7

OBJECTIVE

Earthquake–induced building damage

area, grade of damage, type of damage

European Macroseismic Scale 1998 (EMS98)

Grade 3 – 5 (EMS98)

Grade 1 Grade 2 Grade 3 Grade 4 Grade 5

Page 8: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 8

OBJECTIVE

Earthquake–induced building damage

area, grade of damage, type of damage

European Macroseismic Scale 1998 (EMS98)

Grade 3 – 5 (EMS98)

Individual building,

Roof damage, and

Façade structural damage detection

Building Damage Assessment Map

Grade 1 Grade 2 Grade 3 Grade 4 Grade 5

Page 9: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 9

OBJECTIVE

Earthquake–induced building damage

area, grade of damage, type of damage

European Macroseismic Scale 1998 (EMS98)

Grade 3 – 5 (EMS98)

Individual building,

Roof damage, and

Façade structural damage detection

Building Damage Assessment Map

Grade 1 Grade 2 Grade 3 Grade 4 Grade 5

Evaluate with:

Spectral information in nadir and oblique imagery, and

Geometric information in photogrammetric DSM

Page 10: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 10

Photogrammetric processing

Input imagery

Diff. resolution

Diff. imaging geometry

Manned/unmanned, nadir/oblique

Different resolution

Different imaging geometry

Consequences/challenges

Varying scale within images

Less similarity between images

Occlusions

Complex image overlaps

AUTOMATED PROCESSING MORE DIFFICULT

Page 11: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 11

Photogrammetric processing

Image orientation [Rupnik et al., 2013]

Tie point extraction

Approximate orientation

Direct methods

Orientation built step by step

Careful concatenation necessary

Refined bundle adjustment

Image dense matching [Nex and Remondino, 2012]

Dense but 1 object point per pixel

Noisy Shadows and occlusion

True orthophoto generation

NA

DIR

O

BL

IQU

E

Page 12: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT

Given

Spectral → Geometric information

Photogrammetric DSM

True orthophoto

Goal

Thematic information

Ground

Vegetation

Building

Roof

Facade

12

Earthquake damage assessment – methods

Classification 1/3

implicit

explicit

Page 13: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 13

Earthquake damage assessment – methods

Classification 1/3

Page 14: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 14

Earthquake damage assessment – methods

Page 15: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT

Classification 2/3

Sequential approach

15

Earthquake damage assessment – methods

BUILDING LOCALIZATION

DAMAGE LOCALIZATION

NADIR NADIR

FACADE DAMAGE ASSESSMENT (TO DO)

OBLIQUE

Page 16: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT

BUILDING

VEGETATION HIGH

VEGETATION LOW

GROUND

Classification 2/3

Sequential approach

16

Earthquake damage assessment – methods

BUILDING LOCALIZATION

NADIR

Page 17: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT

BUILDING INTACT

BUILDING DAMAGED

VEGETATION HIGH

VEGETATION LOW

GROUND

Classification 2/3

Sequential approach

17

Earthquake damage assessment – methods

BUILDING LOCALIZATION

DAMAGE LOCALIZATION

NADIR NADIR

XZ = DSM – DEM XSV = SLOPE VARIABILITY OVER A CONSTRAINED REGION FEATURES X

Page 18: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT

verification of harmed entities structural damage assessment by

pointcloud modelling and 2D image processing

Classification 2/3

Sequential approach

18

Earthquake damage assessment – methods

BUILDING LOCALIZATION

DAMAGE LOCALIZATION

NADIR NADIR

FACADE DAMAGE ASSESSMENT (TO DO)

OBLIQUE

Page 19: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT

Classification 3/3

Dataset

Where – San Felice, Italy

When – May 2012

Magnitude – 6.0 MMS

Nadir post – UltraCAM XP, ~5cm GSD, 80/60 overlap

Oblique post – Midas 5, ~10cm GSD (nadir), 70/50 overlap

19

Earthquake damage assessment – results

Page 20: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT

Classification 3/3

20

Earthquake damage assessment – results

B DAMAGED B INTACT

VEGETATION GROUND

Area 1

Page 21: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT

Classification 3/3

21

Earthquake damage assessment – results

B DAMAGED B INTACT

VEGETATION GROUND

Area 2

Page 22: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT

A methodology was presented that uses

- spectral and

- geometric information derived from optical images

in order to classify buildings as intact or damaged in an unsupervised manner;

Adopted features (PNDVI, Z, SLOPE VARIABILITY) with their normaliztion factors are discriminative yet occasionally might prove insufficient thus more feature should be tested and added to the model (possibly through learning);

EMS Grade 4 -5 are detectable with the given approach

Further novelty of the approach is in the unsupervised approach and the use of photogrammetric DSM;

Despite the developments in digital photogrammetry, the DSM produced in an automated manner can be noisy because of occlusion and shadowing effects;

Appropriate filtering techniques can largely mitigate the blunders in DSM

Future works include the use of oblique images and the derived pointclouds to verify the potentially harmed entities as well as evaluate EMS Grade 3 damages

22

Conclusions and future work

Page 23: Automated processing of high resolution airborne images ...rapidmap.fbk.eu/sites/rapidmap.fbk.eu/files/Rapidmap_Tokyo_Rupnik.pdf · 3 E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES

E. Rupnik – HIGH RESOLUTION AIRBORNE IMAGES FOR EARTHQUAKE DAMAGE ASSESSMENT 23

Bibliography

Adams, Stuart M., and Carol J. Friedland, 2011. A survey of unmanned aerial vehicle (UAV) usage for imagery collection in disaster research and management. 9th International Workshop on Remote Sensing for Disaster Response.

Boykov, Y., Veksler, O., Zabih, R., 2001. Fast approximate energy minimization via graph cuts. IEEE Transactions on Pattern Analysis and Machine Intelligence 23 (11), 1222–1239.

Chou, T.-Y., M.-L. Yeh, et al., 2010. Disaster Monitoring and Management by the Unmanned Aerial Vehicle Technology. ISPRS Technical Commission VII Symposium. W. Wagner and B. Szekely. Vienna, Austria, IAPRS. XXXVIII: 6.

Dong, L., and Shan J., 2013. A comprehensive review of earthquake-induced building damage detection with remote sensing techniques. ISPRS Journal of Photogrammetry and Remote Sensing 84, 85-99.

Gerke, M., Kerle, N., 2011. Automatic structural seismic damage assessment with airborne oblique pictometry imagery. Photogrammetric Engineering and Remote Sensing 77 (9), 885–898.

Huber, M. 2010. Evergreen supports UAV team mapping Haitian Relief. Aviation International News. March 2010.

Li, Z., Jiao, Q., Liu, L., Tang, H., & Liu, T., 2014. Monitoring Geologic Hazards and Vegetation Recovery in the Wenchuan Earthquake Region Using Aerial Photography. ISPRS International Journal of Geo-Information 3.1, 368-390.

Nex, F., Remondino, F., 2012: Automatic roof outlines reconstruction from photogrammetric DSM. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. I(3), 257-262.

Rupnik, E., Nex, F., and Remondino, F., 2013. Automatic orientation of large blocks of oblique images. Int. Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. 40(1/W1), 299-304