Mapping the Path to Digital Camera Calibration

Download Report

Transcript Mapping the Path to Digital Camera Calibration

Digital Aerial Imaging Systems:
Current Activities and Issues
American Society for Photogrammetry and Remote
Sensing Annual Meeting,
Baltimore, Maryland
09 March 2005
U.S. Department of the Interior
U.S. Geological Survey
National Center for EROS
Remote Sensing Technologies Group
1
Why Laboratory Calibration Is Not The
Answer For Digital Aerial Systems
Diverse architectures
Integration of cameras with other systems such as IMU
and airborne-GPS equipment
Integration with software processes
The government can neither build a single
instrument to calibrate all systems nor build
multiple instruments to cover every possible
architecture.
U.S. Department of the Interior
U.S. Geological Survey
2
In Situ Methods
In situ methodologies can account for and measure
sources of error in an imaging system.
Geopositional – Use checkpoints with known x, y, and z values
Spatial – Ground Sample Distance and Modulation Transfer
Function (MTF). Estimate MTF via:


Point Spread Functions – point targets
Line Spread Functions – pulse targets or edge targets
Radiometric


Absolute Radiometry
Relative Radiometry
U.S. Department of the Interior
U.S. Geological Survey
3
The Goals of the C2V2 Group
Long-Term: Federal Civil Policies, Standards, and
Guidelines for Digital Aerial Imaging Systems
Short-Term: Remove the Barriers that Prevent the Use of
Digital Aerial Imaging Systems
U.S. Department of the Interior
U.S. Geological Survey
4
In the Interim…
Product Characterization: Evaluate products and
compare results to product specifications.
An interim solution until policies, standards, and guidelines can
be emplaced.
Jointly conducted by USGS and NASA
Operators submit orthoimagery of Stennis Space Center ‘Fee
Area’ using Ground Sample Distances and elevation sources
they would use to fill government contracts.
U.S. Department of the Interior
U.S. Geological Survey
5
Product Characterization: Pro’s
Straightforward implementation
Product specification driven – not constrained by system
architecture
Each test increases experience with these systems and their
datasets; the more we learn, the better our policies, standards, and
guidelines will become
U.S. Department of the Interior
U.S. Geological Survey
6
Product Characterization: Con’s
Cost to both operators and government
Voluntary participation
‘Benchmark’ type test is not an indicator of long-term or
sustained performance.
Orthoimage compilation process can introduce error to
the product that does not originate with the imaging
system.
U.S. Department of the Interior
U.S. Geological Survey
7
Evaluation Components
Geopositional – Compare measured coordinates of
checkpoints in the orthoimagery to surveyed values.
Spatial – Perform a Relative Edge Response analysis on
imagery of edge targets. This is a technique that
estimates the Line Spread Function of an imaging
system.
Radiometric – Deploy special tarps and other
instruments.
U.S. Department of the Interior
U.S. Geological Survey
8
Geopositional
Use checkpoints to assess
imagery
Geodetic Targets
Manhole Covers
U.S. Department of the Interior
U.S. Geological Survey
9
Spatial
Use Edge Targets to evaluate Relative Edge Response (RER)
RER  [ERX (0.5)  ERX (0.5)][ERY (0.5)  ERY (0.5)]
The RER is a geometric mean of
normalized edge response differences
measured in two directions of image
pixels (X and Y) at points distanced
from the edge by -0.5 and 0.5 GSD.
The RER estimates effective
slope of the imaging system’s
edge response because
distance between the points
for which the differences are
calculated is equal to the
GSD.
U.S. Department of the Interior
U.S. Geological Survey
10
Image area selected for spatial
response
measurement in Northing
direction
Radiometric
Absolute correction: Correct radiance or reflectance should be
measured or converted by using the sensor calibration data, the
sun angle and view angle, atmospheric models and ground truth
data.
Relative Correction: Relative correction is to normalize multitemporal data taken on different dates to a selected reference data
at specific time.
Typical techniques:
Adjustment of average and standard deviation values.
Conversion to normalized index: for example the normalized difference
vegetation index (NDVI).
Histogram matching: the histograms per band and/or per sensor are
calculated and the cumulative histogram with cut-offs at 1% is determined,
where y is reference data and x is data to be normalized.
U.S. Department of the Interior
U.S. Geological Survey
11
Current Status
Seven evaluations completed
Applanix DSS 300 operated by Emerge (January 2003) –
Geopositional Only
Space Imaging Digital Airborne Imaging System operated by
Space Imaging (November 2003)
Leica Geosytems ADS40 operated by EarthData International
(November 2002)
IKONOS Satellite operated by Space Imaging (December
2003)
Zeiss/Intergraph Digital Mapping Camera operated by AEROMETRIC, Inc. (February 2004)
M7 Visual Intelligence AirRecon V operated by M7 Visual
Intelligence (December 2004) – Geopositional Only
U.S. Department of the Interior
U.S. Geological Survey
12
Current Status (Continued)
Three evaluations pending
Leica Geosystems ADS40 flown by Northwest Geomatics
(October 2003)
Zeiss/Intergraph Digital Mapping Camera operated by 3001, Ltd.
(2004)
Vexcel UltraCam-D operated by Sanborn (December 2004)
Interest from other sensor operators: DeLorme; Digital Aerial
Solutions; GeoVantage; Horizons, Inc.; Photo Science, Inc.;
Spectrum Mapping; and Titan
U.S. Department of the Interior
U.S. Geological Survey
13
Results
Product
Operator
Sensor
Type
Check
GSD (M)
CE90 (M)
CE95 (M)
RMSENet (M)
Points
Emerge
Applanix DSS300
RGB
0.30
0.48
0.54
0.31
150
Space Imaging
DAIS
RGB
0.50
0.73
0.83
0.48
150
RER
RER Band
Saturated
0.85
Blue
0.76
Green
0.47
Red
0.74
Near IR
Space Imaging
IKONOS
PAN
1.00
2.28
2.53
1.65
41
0.75
Panchromatic
EarthData
Leica Geosystems
ADS40
RGB
0.20
0.43
0.49
0.29
183
0.49
Panchromatic
0.39
Blue
0.55
Green
0.54
Red
0.62
Near IR
0.41
Panchromatic
AERO-METRIC, Inc.
Z/I DMC
PAN
0.15
0.27
0.31
0.18
43
M7 Visual Intelligence
AirRecon V
RGB
0.25
0.64
0.72
0.43
45
Saturated
RGB
0.50
1.11
1.20
0.66
45
Saturated
BOLD =
U.S. Department of the Interior
U.S. Geological Survey
14
Empirical CE Calculation
USGS Specifications for
Orthoimagery
GSD (m) GSD (ft) Map ScaleRMSEx,y RMSEr CE90 CE95
0.15
0.50
2400
0.61
0.86 1.31 1.49
0.25
0.80
4800
1.22
1.73 2.62 2.99
0.30
1.00
8400
2.12
3.00 4.55 5.19
0.50
1.64
9000
3.55
5.02 7.62 8.69
1.00
3.28
12000
4.73
6.70 10.16 11.59
*RMSE and CE values are in meters
Specifications sources:
•1.00M: USGS Orthoimagery Standards
•0.50M, 0.25M, 0.15M: Large Scale Mapping Guidelines
•0.30M: Product Specifications for High Resolution Urban Area
Imagery for NGA
U.S. Department of the Interior
U.S. Geological Survey
15
Results-Specifications Comparison
Operator
Sensor
GSD (m)
0.15
AERO-METRIC, Inc.
DMC
GSD (ft)
CE90
0.50
0.15
0.25
0.80
CE95
1.31
1.49
0.27
0.31
2.62
2.99
M7 Visual Intelligence
AirRecon V
0.25
0.64
0.72
EarthData International
ADS40
0.20
0.43
0.49
4.55
5.19
0.48
0.54
7.62
8.69
0.30
Emerge
DSS300
1.00
0.30
0.50
1.64
M7 Visual Intelligence
AirRecon V
0.50
1.11
1.20
Space Imaging
DAIS
0.50
0.73
0.83
10.16
11.59
2.28
2.53
1.00
Space Imaging
U.S. Department of the Interior
U.S. Geological Survey
IKONOS
1.00
16
3.28
Issues
Diversity: Operator capabilities and products vary; it is
hard to make a meaningful comparison of one
operator’s product to another.
Variety of GSD’s: six-inch to one meter.
Variety of image types: panchromatic, natural color, CIR.
Variety of elevation data sources – USGS DEM’s, LiDAR,
ISTAR, image autocorrelation
Relative Edge Response
Sensitivity: Vulnerable to image re-sampling and image
saturation.
Relevance: What is the significance of the RER value?
U.S. Department of the Interior
U.S. Geological Survey
17