Transcript Project Overview

Project Overview
COMPARISON
BETWEEN AERIAL DIGITAL
ORTHOPHOTO AND
SATELLITE IMAGES
GISDATA D.O.O.
Ivana Lampek Pavčnik,
[email protected]
Integrating knowledge, technology & data into working systems
Project Goals

COMPARISON:
– Quality of geometrical corrections
– Quality of Interpretability
– Time for defining and ordering
– Time for geometric correction
– Price
Description

The focus of the project was to
examine the results of different type
comparisons and discuss the
advantages or disavantages between
aerial and satellite images
FOR MORE INFO...
See the final report and procesed data
Data used in project

Aerial black/white photos
– Area of interest: city Karlovac and
environment
– Area: 42000 m2
– Scale of expose: 1:20000
– Number of frames: 10, spatial resolution=0,5m
– Date:
• For the frame 317: 04. 05. 2000.
• For the frame 2/1: 29. 02. 2000
Data used in project

Color aerial photos
– Area of interest: city Karlovac and
environment
– Area: 27000 m2
– Scale of expose: 1:20000
– Number of frames: 6, spatial resolution=0,5m
– Date:
• May, 2002.
Data used in project

IKONOS satellite images
– Area of interest: city Karlovac and
environment
– Area: 57 000 m2
– Number of frames: 2, spatial resolution=1m
– Date:
• May, 2003.
Technology
Digital
Photogrammetry
for geometrical corrections
Digital Photogrammetry
Goal: Creating Orthos
 Means

– Aerial Triangulation
– Orthorectification
Why ORTHOrectify?
There are geometric errors
associated with satellite images and
aerial photographs
 Errors are caused by:
– Scale Variation
– Sensor Attitude/Orientation
– Internal Sensor Errors
 Orthorectification removes these
errors

Scale Variation
House width = 8m
2 cm
Scale is 1:400
6 cm
Scale is 1:133
Scale varies across the photography
Scale Variation
House width constant (8m), width in photographs varies, therefore scale varies
Differences between aerial
triangulations



Aerial photos> to establish Image
Coordinates
Provided in a Camera Calibration Certificate
Parameters defining this geometry are:
– Focal length
– Radial Lens Distortion
– Principal Point
– Fiducial Coordinates
Image/Focal Plane
Focal Length
Optical Axis
Internal Geometry of satellite

Usually the internal parameters are read
from the image header (SPOT, IRS):
– Focal length
– Principal point Xo, Yo
– Pixel Size
– Number of sensor Columns
 In the IKONOS and Quick Bird case, the
geometry is modeled using rational
polynomials
 User does not need to define these
Flight Line Characteristics for aerial
photos
A block should have at least one pair of images that overlap
IKONOS or Quick Bird images do
not need to have at least one
pair of images that overlap
Acquiring Ground GCP
Coordinates

Coordinates of GCPs in external
orientation can be gathered using various
techniques:
– Using GPS
– From Maps
– From other rectified imagery

Should have X, Y and Z values for
overlapping aerial images
 Should have X, Y for satellite images and Z
values from DEM
The Influence of Quality Estimates
Adjustment process will move points
until the “best solution” is found
Image
Inputted
Standard Deviations
(Measures of Quality)
The points fluctuate with weighted limits as
specified by the standard deviation values
Adjustment takes places in the X,
Y AND Z direction
Ground
Block Residuals
• Block of eight images…
• Image & ground measurements
• Least Squares Adjustment
calculates new points based on
distributing and minimizing
residuals throughout the
ENTIRE block
• There are RESIDUALS
for:
- Each ground point
- Each image point
- Each perspective center
Block Residuals for all data sources

Color Aerial images:
–

mX
0.3239
mZ
0.6507
B&W Aerial images :
– mX
0.4175

mY
0.309
mY
0.4641
IKONOS:
mX
0.3879
mZ
0.4220
mY
0.3687
mZ
DEM -accuracy
RESULTS of geometrical correction:
2
3
1. Pixel in the DEM (Height)
2. Parameters of Interior and Exterior
Orientation
3. In the image, a brightness value is
determined based on the
resampling of surrounding pixels
1
4
Orthographic Projection
The orthographic image is
constructed by resampling
the original image pixels into
their new orthorectified
positions
4. Height, Interior and Exterior
Orientation information and
Brightness Value are used to
calculate equivalent location in the
Ortho Image
Digital terrain model:
- 32 digitized maps with scale 1: 5000
- equidistance: 5m
-summary: 125 878 arcova for generating the surface model
Digital orthophotos


CORRECTED Images as result of
ortorectification process
The software takes each DEM pixel and finds the equivalent
position in the image. A brightness value is calculated
based on the surrounding pixels. This brightness value, the
elevation, the interior orientation and exterior orientation
information is used to calculate the equivalent location on
the ortho image
Quality of interpretability
Automatic interpretation
 Defining the level of

Image Interpretability Rating Scales
Automatic interpretation

Seed properties>
Neighborhood: This option determines which pixels will be considered
contiguous to the seed pixel. Any neighbor pixel that meets all selection
criteria is accepted and thus, itself, becomes a seed pixel.
If four neighbors are searched, then only those pixels above, below, to
the left, and to the right of the seed pixel are considered contiguous.
If eight neighbors are searched then the diagonal pixels are also
considered contiguous.
Geographic Constraints: This group allows you to enter constraints for the
AOI. You can select only one option or use both options.
Area: The maximum size of the AOI
Distance: specifying a distance from the seed pixel.
Spectral Euclidean Distance: The Euclidean spectral distance in digital
number (DN) units on which to accept pixels. The pixels that are accepted will
be within this spectral distance from the mean of the seed pixel.
SED= 10, AREA=5000 pixels
SED= 49, AREA=5000 pixels
SED= 50, AREA=5000pixels
SED= 10, AREA=5000 pixels
SED= 13, AREA=5000 pixels
SED= 10, AREA=5000 p
UNSUPERVISED CLASSIFICATION

ISODATA algorithm to perform an unsupervised
classification. ISODATA stands for "Iterative SelfOrganizing Data Analysis Technique.“

It is iterative in that it repeatedly performs an entire
classification (outputting a thematic raster layer) and
recalculates statistics. "Self-Organizing" refers to the way
in which it locates the clusters that are inherent in the data.

The ISODATA clustering method uses the minimum
spectral distance formula to form clusters. It begins with
either arbitrary cluster means or means of an existing
signature set, and each time the clustering repeats, the
means of these clusters are shifted. The new cluster
means are used for the next iteration.
UNSUPERVISED
CLASSIFICATION

The ISODATA utility
repeats the clustering
of the image until
either:
– a maximum number of
iterations has been
performed, or
– a maximum percentage
of unchanged pixels
has been reached
between two iterations.
Identification of agriculture
Identification of forest
Interpretation into 20 category
IKONOS image
Aerial color image
Defining the level of
Image Interpretability Rating Scales

National Imagery Interpretability Rating
Scale (NIIRS) >
– to define and measure the quality of images
and performance of imaging systems
– NIIRS has been primarily applied in the
evaluation of aerial imagery, it provides a
systematic approach to measuring the quality
of photographic or digital imagery, the
performance of image capture devices, and the
effects of image processing algorithms.
NIIRS 5 [0.75 - 1.2 m GRD]
1m IKONOS images, aerial images 1:40000 i s.r. 21-28μm
Output scale 1:5000
Visible
NIIRS
Distinguish between a
MIDAS and a
CANDID by the
presence of refueling
equipment (e.g.,
pedestal and wing
pod).
Identify radar as vehiclemounted or trailermounted.
Identify, by type, deployed
tactical SSM systems
(e.g., FROG, SS-21,
SCUD).
Distinguish between SS-25
mobile missile TEL
and Missile Support
Vans (MSVS) in a
known support base,
when not covered by
camouflage.
Identify TOP STEER or
TOP SAIL air
surveillance radar on
KIROV-,
SOVREMENNY-,
KIEV-, SLAVA-,
MOSKVA-, KARA-,
Radar
NIIRS
Count all medium
helicopters (e.g.,
HIND, HIP, HAZE,
HOUND, PUMA,
WASP).
Detect deployed TWIN
EAR antenna.
Distinguish between river
crossing equipment
and medium/heavy
armored vehicles by
size and shape (e.g.,
MTU-20 vs. T-62
MBT).
Detect missile support
equipment at an SS-25
RTP (e.g., TEL,
MSV).
Distinguish bow shape and
length/width
differences of SSNS.
Detect the break between
railcars (count
railcars).
Infrared
NIIRS
Distinguish between singletail (e.g., FLOGGER,
F-16, TORNADO) and
twin-tailed (e.g., F-15,
FLANKER,
FOXBAT) fighters.
Identify outdoor tennis
courts.
Identify the metal lattice
structure of large (e.g.
approximately 75
meter) radio relay
towers.
Detect armored vehicles in a
revetment.
Detect a deployed TET
(transportable
electronics tower) at
an SA-10 site.
Identify the stack shape
(e.g., square, round,
oval) on large (e.g.,
greater than 200
meter) merchant ships.
Multispectral
NIIRS
Detect automobile in a
parking lot.
Identify beach terrain
suitable for
amphibious landing
operation.
Detect ditch irrigation of
beet fields.
Detect disruptive or
deceptive use of paints
or coatings on
buildings/structures at
a ground forces
installation.
Detect raw construction
materials in ground
forces deployment
areas (e.g., timber,
sand, gravel).
NIIRS 6 [0.40 - 0.75 m GRD]
0,58 m QUICK BIRD images, aerial photos 1:20000 i 14-28μm
Output scale: 1:2000
Visible
NIIRS
Distinguish between models
of small/medium
helicopters (e.g.,
HELIX A from
HELIX B from
HELIX C, HIND D
from HIND E, HAZE
A from HAZE B from
HAZE C).
Identify the shape of
antennas on
EW/GCI/ACQ radars
as parabolic, parabolic
with clipped comers or
rectangular.
Identify the spare tire on a
medium-sized truck.
Distinguish between SA-6,
SA- I 1, and SA- 17
missile airframes.
Identify individual launcher
covers (8) of vertically
launched SA-N-6 on
SLAVA-class vessels.
Identify automobiles as
Radar
NIIRS
Distinguish between
variable and fixedwing fighter aircraft
(e.g., FENCER vs.
FLANKER).
Distinguish between the
BAR LOCK and SIDE
NET antennas at a
BAR LOCK/SIDE
NET acquisition radar
site.
Distinguish between small
support vehicles (e.g.,
UAZ-69, UAZ-469)
and tanks (e.g., T-72,
T-80).
Identify SS-24 launch triplet
at a known location.
Distinguish between the
raised helicopter deck
on a KRESTA II (CG)
and the helicopter
deck with main deck
on a KRESTA I (CG).
Infrared
NIIRS
Detect wing-mounted stores
(i.e., ASM, bombs)
protruding from the
wings of large
bombers (e.g., B-52,
BEAR, Badger).
Identify individual
thermally active
engine vents atop
diesel locomotives.
Distinguish between a FIX
FOUR and FIX SIX
site based on antenna
pattern and spacing.
Distinguish between
thermally active tanks
and APCs.
Distinguish between a 2-rail
and 4-rail SA-3
launcher.
Identify missile tube hatches
on submarines.
Multispectral
NIIRS
Detect summer woodland
camouflage netting
large enough to cover
a tank against a
scattered tree
background.
Detect foot trail through tall
grass.
Detect navigational channel
markers and mooring
buoys in water.
Detect livestock in open but
fenced areas.
Detect recently installed
minefields in ground
forces deployment
area based on a regular
pattern of disturbed
earth or vegetation.
Count individual dwellings
in subsistence housing
areas (e.g., squatter
settlements, refugee
camps).
Time for defining and ordering

Aerial images in archive: 10-15days

IKONOS in archive: 10-15days
– Min.order=100km2

Quick Bird in archive: 10-15days
– Min.order=64km2
Time for geometric correction

Triangulation:
– Aerial (10 frames)=2,5 days
– IKONOS (2 frames)= 1 day

Ortorectification: the same
 Color matching and mosaic:
– Aerial b/w (10 frames)=3 days
– Aerial color (10 frames)=4 days
– IKONOS color (2 frames) =1 day
Summary:
Aerial color : 7,5 days
Aerial b/w : 6,5 days
IKONOS color : 3 days
PRICE:
Aerial photos (b/w)= 6,53 €/km2
 Aerial photos (color)= 7,84 €/km2
 IKONOS images = 25,80 €/km2
 Quick Bird = 25,80 €/km2

CONCLUSION
Archives
Aerial b/w
Aerial color
Ikonos
Quick
Bird
PRICE
6,53
5
7,84
5
25,8
3
25,8
3
Points
Archives
Aerial b/w
Aerial color
Ikonos
Quick
Bird
Num. Of Frames
12
12
2
2
Time for aerial
triangulations for
defined area:
50 000m2 (days)
2,5
2,5
1
1
Time for orthophotos
ortofoto-a (night)
1
1
1
1
Time for color balance
and mosaic
3
4
1
1
Summary:
6,5
7,5
3
3
Points:
4
3
5
5
Archives
Aerial b/w
Aerial color
Ikonos
Quick
Bird
Num. Of Frames
12
12
2
2
Days for order:
10-15
10-15
10-15
10-15
Points:
4
4
4
4
New images
Aerial b/w
Aerial color
Ikonos
Quick
Bird
Num. Of Frames
12
12
2
2
Days for order:
1month
1 month
Points:
4
4
1-3
mont
hs
3
1-3 month
s
3
Interpretability
Aerial b/w
Aerial color
Ikonos
Quick Bird
Spectral
3
4
5
5
Geometrical
5
5
3
5
Points:
8
9
8
10
Archives
Aerial b/w
Aerial color
Ikonos
Quick
Bird
PRICE
5
5
3
3
TIME for
4
3
5
5
4
4
4
4
Interpret
ability
8
9
8
10
Points:
21
21
20
22
procesing
Time for
order