rEMOTE sENSING AND GIS _11-converted.pdf

RS AND GIS

1. ES 7-Principles of
REMOTE SENSING
AND GEOGRAPHIC
INFORMATION
SYSTEM
Engr. Erwin C. Torio, Ph.D.

2. Remote Sensing Course
Module-11
High Resolution Satellite Imagery (HRSI)

3. Learning Objectives of Module 11
“HRSI”
To learn the state-of-the-art of HRSI
To understand the mapping capability
of HRSI
To learn the prospects of HSRI
applications

4. Outline of Module –11 HRSI
Contents:
1.Road map toward HRSI
2.Characteristics of HRSI
3.Provided products of HRSI
4.Imaging geometry for satellite
line scanners
5.Metric performance of different
satellite sensors

5. 6.Requirements for topographic maps
7.Control points for HRSI
8.2D Geo-positioning
9.3D modeling
10.Applications of HRSI
Contents (2)
QuickBird
©DigitalGlobe

6. Road map toward HRSI
1972: Landsat MSS, 80m
1982: Landsat TM, 30m
1986: Spot, 10m Pan
1991: KVA 1000, 2m
1996: IRS 1C/D, 6m
1999: IKONOS 1, 1m
2000: EROS A1, 1.8m
2001: QuickBird, 0.6m
2002: Spot 5, 2.5m
2003: Orbview, 1m
Landsat TM
IKONOS
QuickBird

7. Launched 1 meter Resolution
Satellites
Satellite IKONOS EROS QuickBird Orbview
Ground 0.82m Pan 1.8m Pan 0.61m Pan 1m Pan
Resolution 3.28m MS 2.5m MS 4m MS
Repeat
cycles
14 days 15 days 20 days 16 days
Launched 1999 2000 2001 2003

Operator
Space
Imaging
West
Indian
Space
Digital
Globe
Orbital
Sciences
© Shunji Murai 2004

8. Characteristics of HRSI
Mapping capability
High geometric fidelity
Highly automated product generation
Potentially very rapid cycle of image
collection to customer delivery
IKONOS
©SpaceImaging

9. Sensor and Mission Parameters
Item IKONOS QuickBird
Focal Length 10m 9m
Altitude 680km 450km
No of pixel/l 13,800 27,500
FOV 0.93deg. 2.1deg.
Resolution 0.82m 0.61m
Revisit 1~3 days 1~3.5 days
Coverage 11x11km 16.5x16.5km

10. Provided Products of HRSI
Geocoded Image: Pan, MSS, Pan sharpened
Orthoimage: corrected for topography
DSM/DEM: from stereo imagery
Contour line map: to be generated from
DEM
Land cover map: auto/semi-auto
Overlay on GIS: background image
3D landscape: animation

11. Downtown Tokyo
Example of IKONOS Image
1meter
resolution
Panchromatic
image

©Copyright
Japan Space
Imaging

12. Example of IKONOS Image
4m multi-
spectral
image of
Mt. Everest
©SpaceImaging

13. Pan-sharpened Imagery
4-Meter Multispectral
Image
1-Meter Panchromatic
Image
1-Meter Pan-Sharpened
Image
IKONOS example

14. 13
Saitama, Japan
Example of
IKONOS
Image
Pan sharpened
1meter
resolution

©Copyright
Japan Space
Imaging

15. Example of QuickBird Image
Diet in Tokyo ©Copyright DigitalGlobe

16. Example of
QuickBird
Image
Pan sharpened
0.7 meter
resolution

©DigitalGlobe
Yokohama
Japan

17. QuickBird versus IKONOS
QuickBird 70cm Pan IKONOS 1m Pan
Tennis Centre, Melbourne © Clive Fraser

18. Conducive to
building feature
measurement
Comparative Resolution
Aerial Photography (1:15K) Ikonos Pan Ikonos Pan-Sharpened
© Clive Fraser

19. 18
Comparison of Scene Coverage
QuickBird – 16.5km
Eros 1A – 12.5km
Landsat 7 Image of Washington D.C
Ikonos – 11km
© DigitalGlobe

20. 19
Orthoimage overlaid by Vector Layers
Automated
generation
of ortho-
imagery,
but no fully
automated
feature
extraction
© Clive Fraser

21. 20
Pan-sharpened IKONOS 1m Ortho-image
draped over a DTM
© Clive Fraser

22. Potential of HRSI
Mapping ( to 1:5K-1:10K)
GIS
Provision of DTMs
Automated Feature Extraction
Fused Data Products
Visualization & 3D Fly Through

23. Mapping Capability of HRSI
1m ground resolution with 0.3 pixel
pointing accuracy will provide 1:10,000
line drawing map with the contour
interval of 2.5-5m
Background image map will be possible
at the scale of 1:5,000 including
orthoimage

24. Imaging geometry for satellite
line scanners
Strip
Ground
Swath
Along track
motion
X Y
H
Detector
f Optics
FOV
Different exterior
orientation for every line
Transformation from
Image to Object Space
(x,y)  (X,Y,Z)
Where
•x is the line number (function
of time)
•Object Points (X,Y,Z)
j, j = 1, n
•Images(X
ct,Y
ct,Z
ct)
ik , k = 1,
number of scan lines; i =1,m
•Image Points: (x,y)
ijk
© Clive Fraser

25. Concept of Stereo or Multi-stations

XY
© Clive Fraser
= S 
xy

Z = (H/B) S 
px
S = H/c
(alititude/focal
length)
px = parallax meas.
B = base length

26. Metric Performance of
Different Satellite Sensors
Satellite Resolution Map scale
Landsat 30 m 1: 250K
SPOT 10 m 1: 100K
IRS 6 m 1: 50K
IKONOS below 1 m 1: 10K
QuickBird below 1 m 1: 10K

27. Requirements for Topographic
maps
Scale Planimetric Vertical Contour int.
1:50K 10m 3-6m 10-20m
1:25K 5m 3m 10m
1:10K 2m 1.7m 5m
1: 5K 1m 0.3-0.6m 1-2m

28. Control Points for HRSI
Round target with more than 5 pixels in
diameter
Pointing accuracy: 0.05-0.1 pixel
Required Accuracy
 Scale Resolution GCP’s Accur.
1:50K 5m 1.2m
1:25K 2.5m 1m
1:10K 1m 0.5m
1:5K 0.5m 0.1m

29. Control Points: roundabout center
©Copyright Clive Fraser

30. Best-fitting ellipse to edge points in both object & image space
Least-squares template matching
©Copyright Clive Fraser

Examples of Roundabout Targets

31. XY RMS discrepancies in pixels
2D Geo-positioning
Object point XYZ coordinates ‘rectified’
to a ‘projection plane’ based on satellite
position to remove height effects

Image; points
Similarity Affine Projective
X Y XY X Y XY X Y XY
Nadir; 31 0.27 0.33 0.30 0.26 0.32 0.29 0.23 0.30 0.27
‘Left’ steo;26 0.35 0.39 0.37 0.33 0.37 0.35 0.33 0.36 0.35
‘Right’ steo;28 0.39 0.37 0.38 0.38 0.36 0.37 0.39 0.35 0.37
© Clive Fraser

32. 3D Modeling
Stereo imagery needed
Photogrammetric theory unnecessary due
to very narrow FOV (less than 2 degree)
Two methods developed
Affine transformation with 4 parameters
Rational function with 80 parameters
which are provided by HRSI distributor

33. Affine Transformation
x1 = a1X + a2Y + a3Z + a4
y1 = a5X + a6Y + a7Z + a8
x2 = b1X +b2Y + b3Z + b4
y2 = b5X + b6Y + b7Z + b8
x, y: image coordinate
X,Y,Z: ground coordinate

34. Accuracy Check by Affine Model
© Clive Fraser

GCPs

RMS of xy
residuals
(pixels)
Standard err.
(m)
RMS
discrepancies at
checkpoints(m)
xy z Sxy Sz
4 0.14 0.51 1.02 0.49 0.76
6 0.15 0.43 0.88 0.43 0.74
8 0.16 0.40 0.80 0.43 0.74
Example of 2-image IKONOS configurations

35. Rational Function
P
i 4 ( X ,Y ,Z ) j ij
ij P
i 2 ( X ,Y ,Z ) j
P
i 3 ( X ,Y ,Z )
j
P
i1 ( X ,Y ,Z )
j
y 
x 
20 1 19
Z
3
X
2
Z  d  d
2Y  d
3 X  d
4 Z  ...  d P
i 4 ( X ,Y , Z )
j  d
i 3 j 1 2 3 4 19 20
P ( X ,Y , Z )  c  c Y  c X  c Z  ...  c X
2
Z  c Z
3
P ( X ,Y , Z )  a  a Y  a X  a Z  ...  a X
2
Z  a Z
3
i1 j 1 2 3 4 19 20
P ( X ,Y , Z )  b  b Y  b X  b Z  ...  b X
2
Z  b Z
3
i 2 j 1 2 3 4 19 20
where
x
ij , y
ij
normalized image
coordinate
object point coordinate
(normalized lat. Long.
&height)
X,Y,Z

36. Bias Compensation
Each HRSI has always bias of different
size
For Rational Function model, at least a
GCP is needed for bias compensation
80 parameters provided by HRSI
distributor are not compensated for bias

37. Melbourne Ikonos Test Field
Univ. of Melbourne
©Copyright Clive Fraser
3-fold image coverage
7km x 7km area (Dh <
100m)
40 GPS surveyed GCPs
19 building control pts.
sub-pixel, multi-
measurements to image
features
2D & 3D point
determination tests
‘planes of control’ for 2D
tests
building extraction tests

38. 37
Object XYZ Coordinates Transformed
to Pixel Coordinates via Ikonos RPCs
Image Mean
dx
Mean
dy
dx dy
L. Steo. 29.0 16.1 0.41 0.47
Nadir 69.7 -20.6 0.43 0.49
R. Steo. 28.1 16.6 0.46 0.48
Mean and standard error
values for RPC bias-induced
image point perturbations
dx and dy. Units are pixels.
Image coordinate biases (perturbations dx and dy) in RPC
orientation for the stereo and nadir images
Left Nadir Right
©Copyright Clive Fraser
© Clive Fraser

39. Computed Image Coordinate Biases
and RMS Values of Checkpoint
Discrepancies for Ikonos Stereo
Computed image coord.
Biases (pixels)
RMS of checkpoint coord.
(meters)
x y Sx Sy Sz
1 L 27.7 16.7 0.64 0.59 0.90
R 28.4 16.1
2 L 27.7 16.3 0.59 0.61 0.90
R 28.3 15.7
4 L 27.9 16.4 0.59 0.53 0.92
R 28.6 15.8
6 L 28.0 16.4 0.62 0.46 0.90
R 28.8 15.9


No. of
GCPs

40. ©Copyright Clive Fraser

Bias Corrected RPCs
RPCs
(provided)
RMS ~ 4–70 (m)
RPCs
(provided)
RMS ~ 4–70 (m)
RPCs
(corrected)
RMS < 1 (m)
RPCs
(corrected)
RMS < 1 (m)
Suited for immediate application with standard DPWs!

41. Advantages of HRSI
Effective processing with wide coverage
Frequent data acquisition
Good image quality
Simple 3D modeling
Access possibility (high mountain, pole,
boundary etc.)
Less GCPs
No special skill

42. Disadvantages of HRSI
High cost
Cloud coverage
Limited resolution
Fixed frequency and acquisition time
Less experience

43. Applications to Forestry
Broad-leaved forest




Coniferous forest




Broad-leaved forest
Forestry
Classification
of forest tree
© DigitalGlobe. QuickBird image of Kumamoto, Japan

44. Red:Grown-up Crop, Black:Water, White:Crop land

Application to Agriculture
Agriculture
Monitoring
© DigitalGlobe.
False Color True Color

45. 44
Application to Facility Management
QuickBird
image

© DigitalGlobe.

46. Application to Environmental Monitoring
With QuickBird 2.4 m
multispectral
imagery it is possible
for detection of
environmental
problems.
With QuickBird 2.4 m
multispectral
imagery it is possible
for detection of
environmental
problems.
Damaged
Vegetation
from Possible
Oil Leak
© DigitalGlobe

47. Application to 3D City Modeling
3D Model of University of Melbourne Campus
from Ikonos 1m B&W Stereo
© Clive Fraser

48. Comparison of 3D Modeling from
Aerial Photo and HRSI
Aerial Photography (1:15K) Ikonos 1m Stereo Imagery
© Clive Fraser

49. Thank you !
“QuickBird”