Overview Overview Objective Objective Building Inventory Data - - PDF document

▶
overview overview
SMART_READER_LITE
LIVE PREVIEW

Overview Overview Objective Objective Building Inventory Data - - PDF document

Apr 10, 2024 342 likes •493 views

Infrastructure Inventory Compilation Infrastructure Inventory Compilation Using Single High- -Resolution Satellite Resolution Satellite Using Single High Images Images Pooya Sarabandi 1 1 , Beverly Adams , Beverly Adams 2 2 , , Pooya

slide-1
SLIDE 1

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Infrastructure Inventory Compilation Infrastructure Inventory Compilation Using Single High Using Single High-

  • Resolution Satellite

Resolution Satellite Images Images

Pooya Sarabandi Pooya Sarabandi1

1, Beverly Adams

, Beverly Adams2

2,

, Anne S. Kiremidjian Anne S. Kiremidjian1

1, Ronald T. Eguchi

, Ronald T. Eguchi3

3

1 1Stanford University, CA, USA

Stanford University, CA, USA

2 2ImageCat Inc., London, UK

ImageCat Inc., London, UK

3 3ImageCat Inc., CA,USA

ImageCat Inc., CA,USA

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Overview Overview

  • Objective

Objective

  • Building Inventory Data Structure

Building Inventory Data Structure

  • Why Building Inventory?

Why Building Inventory?

  • Available Building Inventory Databases

Available Building Inventory Databases

  • Why Remote Sensing?

Why Remote Sensing?

  • Proposed Algorithm

Proposed Algorithm

  • Sensor Models

Sensor Models

  • Rational Function Model

Rational Function Model

  • Image Acquisition Geometry

Image Acquisition Geometry

  • 3D Reconstruction Algorithm

3D Reconstruction Algorithm

  • Measurement Error

Measurement Error

  • Implementation (MIHEA)

Implementation (MIHEA)

  • Results and Calibration

Results and Calibration

slide-2
SLIDE 2

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Objective Objective

To develop a semi To develop a semi-

  • automated method for

automated method for spatial and structural information extraction spatial and structural information extraction from single satellite images to be used in from single satellite images to be used in building inventories building inventories

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Overview Overview

  • Objective

Objective

  • Building Inventory Data Structure

Building Inventory Data Structure

  • Why Building Inventory?

Why Building Inventory?

  • Available Building Inventory Databases

Available Building Inventory Databases

  • Why Remote Sensing?

Why Remote Sensing?

  • Proposed Algorithm

Proposed Algorithm

  • Sensor Models

Sensor Models

  • Rational Function Model

Rational Function Model

  • Image Acquisition Geometry

Image Acquisition Geometry

  • 3D Reconstruction Algorithm

3D Reconstruction Algorithm

  • Measurement Error

Measurement Error

  • Implementation (MIHEA)

Implementation (MIHEA)

  • Results and Calibration

Results and Calibration

slide-3
SLIDE 3

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Building Inventory Data Structure Building Inventory Data Structure

  • Address or Lon/Lat

Address or Lon/Lat

  • Age or year of construction

Age or year of construction

  • Height or number of stories

Height or number of stories

  • Footprint Area

Footprint Area

  • Structural type

Structural type

  • Occupancy type

Occupancy type

  • Roof type

Roof type

  • Cladding

Cladding

  • Shape irregularity

Shape irregularity

  • Height irregularity

Height irregularity

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Overview Overview

  • Objective

Objective

  • Building Inventory Data Structure

Building Inventory Data Structure

  • Why Building Inventory?

Why Building Inventory?

  • Available Building Inventory Databases.

Available Building Inventory Databases.

  • Why Remote Sensing?

Why Remote Sensing?

  • Proposed Algorithm

Proposed Algorithm

  • Sensor Models

Sensor Models

  • Rational Function Model

Rational Function Model

  • Image Acquisition Geometry

Image Acquisition Geometry

  • 3D Reconstruction Algorithm

3D Reconstruction Algorithm

  • Measurement Error

Measurement Error

  • Implementation (MIHEA)

Implementation (MIHEA)

  • Results and Calibration

Results and Calibration

slide-4
SLIDE 4

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Why Building Inventory? Why Building Inventory?

Vital information to: Vital information to:

  • Decision Makers

Decision Makers

  • Urban Planners

Urban Planners

  • Disaster Response

Disaster Response

  • Loss Estimation

Loss Estimation

  • Portfolio management

Portfolio management

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Overview Overview

  • Objective

Objective

  • Building Inventory Data Structure

Building Inventory Data Structure

  • Why Building Inventory?

Why Building Inventory?

  • Available Building Inventory Databases.

Available Building Inventory Databases.

  • Why Remote Sensing?

Why Remote Sensing?

  • Proposed Algorithm

Proposed Algorithm

  • Implementation (MIHEA)

Implementation (MIHEA)

  • Results and Calibration

Results and Calibration

slide-5
SLIDE 5

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Available Inventory Databases Available Inventory Databases

Sources: Sources:

  • Tax Assessors files

Tax Assessors files

  • Government files (e.g.

Government files (e.g. FEMA, GSA FEMA, GSA … …) )

  • Sanborn maps (US only)

Sanborn maps (US only)

  • Real estate files

Real estate files

  • Insurance portfolios

Insurance portfolios

Problems: Problems:

  • Not always available in

Not always available in digital format digital format

  • Incomplete for many

Incomplete for many attributes attributes

  • Unavailable for many

Unavailable for many regions (e.g. developing regions (e.g. developing countries) countries)

  • Updating only some

Updating only some information information

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Overview Overview

  • Objective

Objective

  • Building Inventory Data Structure

Building Inventory Data Structure

  • Why Building Inventory?

Why Building Inventory?

  • Available Building Inventory Databases

Available Building Inventory Databases

  • Why Remote Sensing?

Why Remote Sensing?

  • Proposed Algorithm

Proposed Algorithm

  • Implementation (MIHEA)

Implementation (MIHEA)

  • Results and Calibration

Results and Calibration

slide-6
SLIDE 6

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Why Remote Sensing? Why Remote Sensing?

  • Remote and hard to reach locations

Remote and hard to reach locations

  • Digital formats

Digital formats

  • Can augment missing information

Can augment missing information

  • Large Coverage

Large Coverage

  • Frequent updates

Frequent updates

  • Eventually will be more cost effective than

Eventually will be more cost effective than land survey land survey

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Building Inventory Data Structure Building Inventory Data Structure

  • Address or Lon/Lat *

Address or Lon/Lat *

  • Age or year of construction

Age or year of construction

  • Height or number of stories*

Height or number of stories*

  • Footprint Area*

Footprint Area*

  • Structural type

Structural type

  • Occupancy type

Occupancy type

  • Roof type*

Roof type*

  • Cladding*

Cladding*

  • Shape irregularity*

Shape irregularity*

  • Height irregularity*

Height irregularity* * * -

  • information that can be obtained from imagery

information that can be obtained from imagery

slide-7
SLIDE 7

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Overview Overview

  • Objective

Objective

  • Building Inventory Data Structure

Building Inventory Data Structure

  • Why Building Inventory?

Why Building Inventory?

  • Available Building Inventory Databases.

Available Building Inventory Databases.

  • Why Remote Sensing?

Why Remote Sensing?

  • Proposed Algorithm

Proposed Algorithm

  • Sensor Models

Sensor Models

  • Rational Function Model

Rational Function Model

  • Image Acquisition Geometry

Image Acquisition Geometry

  • 3D Reconstruction Algorithm

3D Reconstruction Algorithm

  • Measurement Error

Measurement Error

  • Implementation (MIHEA)

Implementation (MIHEA)

  • Results and Calibration

Results and Calibration

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Sensor Models Sensor Models

  • Physical Sensor Model

Physical Sensor Model

  • Sensor dependent

Sensor dependent

  • Parameters have physical significant

Parameters have physical significant

  • Parameters are statistically uncorrelated

Parameters are statistically uncorrelated

  • Rigorous and not always available to users

Rigorous and not always available to users

  • Generalized Sensor Model

Generalized Sensor Model

  • Generic

Generic

  • Sensor independent

Sensor independent

  • Real

Real-

  • time computation

time computation

slide-8
SLIDE 8

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Rational Function Models (RFMs) Rational Function Models (RFMs)

  • Describes the image

Describes the image-

  • to

to-

  • ground relationships.

ground relationships.

  • Generalization of

Generalization of polynomial models (ratio polynomial models (ratio

  • f two polynomial
  • f two polynomial

functions) functions)

1 2 3 4

( , , ) ( , , ) ( , , ) ( , , )

n n n n n n n n n n n n n n

f h r f h f h c f h φ λ φ λ φ λ φ λ = =

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

RFMs cont RFMs cont’ ’d d

Rational Polynomial Rational Polynomial Coefficients (RPCs) Coefficients (RPCs)

3 3 3 1 1 1 i j k ijk i j k

f a h φ λ

= = =

=∑∑∑

1 2 3 4 5 2 2 2 6 7 8 9 10 3 2 2 11 12 13 14 2 3 2 2 15 16 17 18 2 3 19 20

f a a a a h a a h a h a a a h a h a a a h a a a h a h a h a h λ φ λφ λ φ λ φ φλ λ λφ λ λ φ φ φ λ φ = + + + + + + + + + + + + + + + + + + +

slide-9
SLIDE 9

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Approximate image acquisition geometry and satellite Approximate image acquisition geometry and satellite

  • rientation can be described by sensor
  • rientation can be described by sensor’

’s s elevation elevation and and azimuth azimuth angles angles

Image Acquisition Geometry Image Acquisition Geometry

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Height Metrology Height Metrology

  • Image coordinates for the corner of a building at ground level

Image coordinates for the corner of a building at ground level ( (r rground

ground,

, c cground

ground ), and its corresponding roof

), and its corresponding roof-

  • point coordinates

point coordinates ( (r rroof

roof ,

, c croof

roof )

)

  • Sensor

Sensor’ ’s collection azimuth angle s collection azimuth angle

2 2 * *

) ( ) ( H ) cos( H H

roof ground roof ground

c c r r GSD − + − × = = β

slide-10
SLIDE 10

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

3D Reconstruction Algorithm 3D Reconstruction Algorithm

) , , ( ) , , ( ) , , ( ) , , (

2 4 2 3 2 2 2 1

h f h f c h f h f r

roof roof

λ φ λ φ λ φ λ φ = = ) , , ( ) , , ( ) , , ( ) , , (

1 4 1 3 1 2 1 1

h f h f c h f h f r

ground ground

λ φ λ φ λ φ λ φ = =

) (

  • )

, , ( ) , , (

  • )

, , ( ) , , (

  • )

, , ( ) , , (

  • )

, , ( ) , , (

1 2 2 4 2 3 2 2 2 1 1 4 1 3 1 2 1 1

= − − = = = = H h h c h f h f r h f h f c h f h f r h f h f

r r g g

λ φ λ φ λ φ λ φ λ φ λ φ λ φ λ φ

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

  • Homogeneous

Homogeneous

  • Nonlinear

Nonlinear

  • Over

Over-

  • determined

determined Solution: Trust Solution: Trust-

  • Region Dogleg Method

Region Dogleg Method Starting point for the iterative solution: by linearizing RFM Starting point for the iterative solution: by linearizing RFM

3D Reconstruction Algorithm 3D Reconstruction Algorithm -

  • cont

cont’ ’d d

) (

  • )

, , ( ) , , (

  • )

, , ( ) , , (

  • )

, , ( ) , , (

  • )

, , ( ) , , (

1 2 2 4 2 3 2 2 2 1 1 4 1 3 1 2 1 1

= − − = = = = H h h c h f h f r h f h f c h f h f r h f h f

r r g g

λ φ λ φ λ φ λ φ λ φ λ φ λ φ λ φ

4 2 4 3 2 1 2 4 3 2 1 3 2 4 3 2 1 2 4 3 2 1 2 1 4 3 2 1 1 4 3 2 1 1 1 4 3 2 1 1 4 3 2 1

, , ε φ λ φ λ ε φ λ φ λ ε φ λ φ λ ε φ λ φ λ + + + + + + + = + + + + + + + = + + + + + + + = + + + + + + + = h d d d d h c c c c c h b b b b h a a a a r h d d d d h c c c c c h b b b b h a a a a r

r r g g

b A = ⋅

  • x

⎥ ⎥ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎢ ⎢ ⎣ ⎡ − − − − = ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ ⎡ = ⎥ ⎥ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎢ ⎢ ⎣ ⎡ − − − − − − − − − − − − =

1 1 1 1 1 1 1 1 * 2 * 1 * * 3 3 3 3 2 2 3 3 3 3 2 2 3 3 3 3 2 2 3 3 3 3 2 2

, , d c c b r a d c c b r a b h h d d c d d c d d c a b r a b r a b r d d c d d c d d c a b r a b r a b r A

r r g g r r r r r r g g g g g g

λ φ

  • x
slide-11
SLIDE 11

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Measurement Error Measurement Error

  • Image dependent (shadows and obstacles along the line of sight)

Image dependent (shadows and obstacles along the line of sight)

  • User (operator) dependent

User (operator) dependent Pixel H

σ β σ ⋅ ⋅ = ) sec( 2

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Overview Overview

  • Objective

Objective

  • Building Inventory Data Structure

Building Inventory Data Structure

  • Why Building Inventory?

Why Building Inventory?

  • Available Building Inventory Databases.

Available Building Inventory Databases.

  • Why Remote Sensing?

Why Remote Sensing?

  • Proposed Algorithm

Proposed Algorithm

  • Sensor Models

Sensor Models

  • Rational Function Model

Rational Function Model

  • Image Acquisition Geometry

Image Acquisition Geometry

  • 3D Reconstruction Algorithm

3D Reconstruction Algorithm

  • Measurement Error

Measurement Error

  • Implementation (MIHEA)

Implementation (MIHEA)

  • Results and Calibration

Results and Calibration

slide-12
SLIDE 12

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

MIHEA MIHEA

(Mono (Mono-

  • Image Height Extraction Algorithm)

Image Height Extraction Algorithm)

  • Image

Image Processing Processing Package Package

  • Compatible

Compatible with IKONOS with IKONOS and QuickBird and QuickBird camera models camera models

  • Export results

Export results to database to database

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

MIHEA cont MIHEA cont’ ’d d

slide-13
SLIDE 13

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Overview Overview

  • Objective

Objective

  • Building Inventory Data Structure

Building Inventory Data Structure

  • Why Building Inventory?

Why Building Inventory?

  • Available Building Inventory Databases.

Available Building Inventory Databases.

  • Why Remote Sensing?

Why Remote Sensing?

  • Proposed Algorithm

Proposed Algorithm

  • Sensor Models

Sensor Models

  • Rational Function Model

Rational Function Model

  • Image Acquisition Geometry

Image Acquisition Geometry

  • 3D Reconstruction Algorithm

3D Reconstruction Algorithm

  • Measurement Error

Measurement Error

  • Implementation (MIHEA)

Implementation (MIHEA)

  • Results and Calibration

Results and Calibration

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Results and Calibration Results and Calibration

  • Sensor: QuickBird Image

Sensor: QuickBird Image

  • Test Area: City of London

Test Area: City of London

  • Date: July 28, 2002

Date: July 28, 2002

  • Off

Off-

  • nadir viewing angle: 24.6

nadir viewing angle: 24.6o

  • Validation sources:

Validation sources: LiDAR LiDAR and and independently derived survey independently derived survey data data

  • Validation set: 23 buildings

Validation set: 23 buildings selected from 3D model selected from 3D model

50 100 150 200 250 0.5 1 1.5 2 2.5 3 3.5 4 Height (meter) Frequency (number of buildings) Histogram of height distribution LiDAR vs. MIHEA y = 0.6287x + 27.477 R2 = 0.9446

50 100 150 200 250 50 100 150 200 250 300

Height (m) - MIHEA Height (m) - LiDAR Data

Survey Data vs. MIHEA

y = 1.5614x - 16.995 R2 = 0.9542

50 100 150 200 250 300 20 40 60 80 100 120 140 160 180 200

Height(m) - MIHEA Height(m) - Survey Data

slide-14
SLIDE 14

3 3rd

rd International Workshop on Remote Sensing Technologies and Disas

International Workshop on Remote Sensing Technologies and Disaster Response ter Response Chiba, Japan, September 12 ~ 13, 2005 Chiba, Japan, September 12 ~ 13, 2005

ImageCat, Inc.

Questions? Questions?