1 INDIAN SPACE RESEARCH ORGANISATION Evolution of Photogrammetric models in Data Products for ISRO...
-
Upload
dominique-goodrich -
Category
Documents
-
view
223 -
download
0
Transcript of 1 INDIAN SPACE RESEARCH ORGANISATION Evolution of Photogrammetric models in Data Products for ISRO...
1
INDIAN SPACE RESEARCH ORGANISATION
Evolution of Photogrammetric models in Data Products for ISRO RS missions
Presented during World Geo Spatial Forum- 2011
Pradeep K Srivastava,Deputy Director, Signal and Image Processing,
Space Applications Centre (ISRO)Ahmedabad, India
2
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
STEREO STRIP TRIANGULATION (SST) APPROACH
Basically SST is performed over a stereo pair of Cartosat-1 image segment by employing photogrammetric bundle adjustment technique.
InputsSST requires a primary GCP library data with accuracy better than 1 m. Major tasks under SST are
(i) Generation of data base of Triangulated Control Points (TCPs) in stereo mode
(ii) DEM generation over a strip, and(iii) Improvement of image orientation parametersComponentsTCP IdentificationGCP IdentificationRigorous Imaging Model
3
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
Conjugate point identification
Identification of tie points or conjugate points is done in an automatic way by stereo image matching that uses Hierarchical Matching technique.Image Pyramids are formed by sub-sampling the original image at various scales (levels of hierarchy)Selection of optimal number of pyramids depends on the viewing angle of the stereo pair used and terrain undulations.The match points obtained in the last level are the conjugate points used for DEM generation.The process is tuned to identify, one conjugate point for every 100 m * 100 m region on ground.
4
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
TCP identification
TCPs are image points, but they may not be available on a topographic map, like in the case of GCPs. The generation of TCP database involves
• Feature detection to extract candidate • Matching to find the location• Computation of ground coordinates• Blunder detection • Human visual confirmation
Approximately one TCP per square kilometer in Cartosat-1 stereo strip of 500 km length is selected.
5
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
Rigorous Imaging Model
•Cartosat-1 rigorous imaging model is based on photogrammetric collinearity conditions.
•System knowledge over 52 s gap between two cameras in sighting a common GCP is utilised
•Model uses onboard star sensors’ measurement for attitude and GPS based state vector information for orbit.
1. Inertial co-ordinate system(ECI) :2. Geocentric (Greenwich) system, Earth Centered Rotating (ECEF) 3. Local Orbital co-ordinate system: 4. Spacecraft body co-ordinate system: 5. Image co-ordinate system:
(φ, λ, h) Object point in geodetic Everest (Ellipsoid) Frame of reference
through co-ordinate transformation (Spherical to Cartesian)
(X, Y, Z) in Geocentric Cartesian co-ordinate system (ECEF Everest ellipsoid)
through 7 parameter datum conversion
X, Y, Z in WGS 84 (ECEF Greenwich frame)
rotation through sidereal angle
X, Y, Z in Geocentric inertial system(ECI) transformation through orbital elements
X, Y, Z in local orbital co-ordinate system through attitude angles
X, Y, Z in Spacecraft body co-ordinate system tilt angles
X, Y, Z in Focal Plane or Image Plane Coordinate system
and interior orientation Point in image co-ordinate system s,p (scanline and pixel)
The relation between the image and the object co-ordinate systems is expressed by a 3x3 orthogonal matrix designated by M. The nine elements of M are functions of the orientation parameters.
in which v is the vector in x, y, z system and V is the same vector in the X, Y, Z system.
Modified collinearity condition equations for spaceborne imagery
where (x0, y0, -f) is the photo co-ordinate of principle point, (xa, ya) is the photo co-ordinate of an image point , (XA, YA, ZA) is the object space co-ordinate of the same image point and (XL, YL, ZL) is the perspective center co-ordinate in object space co-ordinate system. k is the scale factor that is equal to the ratio of the length of a to the length of A in figure 2. M is the rotation (orientation) matrix from one co-ordinate system to the other, which is a function of exterior orientation parameters.
11
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
GCP identification
•GCPs are the features on ground whose precise ground coordinates are known and which are identifiable on the image.
•A database having well distributed network of ground control points over Indian landmass with ground coordinates obtained using differential GPS and image chips for manual/auto correlation purpose is used for operational use.
In-flight geometric calibration
• Conventional• In-flight calibration successfully
completed for Cartosat-1, Cartosat-2 & Cartosat-2A as part of initial phase exercises/ normal phase to improve the system level performance
• Using Rigorous imaging sensor model techniques with the help of ‘ Imaging sensor data (control points)’
• Achieved accuracies
Satellite Original system Level
(RMS in meters)
Improved thro’ ln-flight calibration
(RMS in meters)
Catosat-1 700 – 1000 m Around 100 m for both Fore and Aft
Cartosat-2
1300 m Better than 50 m
Cartosat-2A
500 – 700 m Better than 70 m
New Techniques• New techniques studied and carried
out R&D exercises for utilising only a set of imaging sensors’ information for Cartosat-1
• Promising results obtained without Control points or a few control points required for validation
• Two Imaging sensors (stereo) from the same orbit of Cartosat-1 are used to derive pseudo attitude
• Platform biases are estimated using only image points or with minimum controls
- Photogrammetry Co-planarity Approach (with/without controls) - Photogrammetry Line based Resection method (no controls) - Accuracy achieved better than 10 pixels for Cartosat-1 - Further study and R&D efforts are on for Chandrayaan-1
National DEM from CARTOSAT-1
• DEM – 1/3 arc-sec posting (~10m)• OrthoImage – 1/12 arc-sec resolution co-
registered (0.1 pixel registration accuracy)• 15 m (CE-90) plannimetric accuracy• 8 m (LE-90) height accuracy• Density of match points ~ 4% using aspect-
based correlation, concensus based outlier identification
• Automatic breakline identification/ incorporation through TIN modeling
DEM generated for Himalayan region using new image matching algorithm
LEGEND SSTS run using GCPS collected from neighboring segments Not processed Outside India Completed
Note : SSTS in the southern India generated using GCPL library (pink color in this region represents non availability of cloud free data )
DEM Status on 27-12-2010
Perspective view of Sonamarg
DEM REPRESENTATION OF SONAMARG AREA IN SRINAGAR DISTRICT
• Four software packages for Chandrayaan-1 Data Processing are operationalised at ISSDC viz.,
– Quick Look Display (QLD) for TMC and HySI with near real time and offline options
– Level-0/1 data processing of TMC and HySI with PDS generation
– Data Archival in PDS standards for all the payload data
– Browse and data visualisation • First day products and many special
products are generated from TMC/HySI and MIS sensors
• Payload Operations Centre (POC) for TMC and HySI is setup at SIPA/SAC for generating higher level products like Lunar DEM, Lunar Atlas and mosaics. This also acts as a gateway for HeX data to PRL.
• A high bandwidth data link is established between SAC and ISSDC and the available raw data has been transferred to SAC-POC
Chandrayaan-1 Data Processing• L-1 DPGS s/w at ISSDC - Operations established in 3 identical chains - No. of data sets processed successfully at ISSDC : TMC – 949 (1071) , HySI - 1071 (1158)
•Active archive for TMC/HYSI/LLRI & AO payloads available at ISSDC
•Long Term Archive (LTA) will be ready by end of 2010
•Data Dissemination of TMC & HYSI from active archive available at ISSDC
• DEM and Lunar Atlas generation at SAC- POC
• Using RPCs & COTS software, DEM generation activity from TMC is on using COTS package customisation. Indigenous software is being developed .
Coordinate system for moon
02 Feb 2010 ISG 2010- Planetary Geomatics: An introduction
0°
TMC NEAR SIDE COVERAGE TMC 90 degree EAST COVERAGE
TMC FAR SIDE COVERAGE TMC 90 degree WEST COVERAGE
90°0°
180°270°
Chandrayaan-1 TMC & HySI Images
Fore Nadir AftFore
Nadir
Aft
TMC ImageOrbit: 1936DOP: 17-04-09
HySI HySI
Band:16-24-32 Band:49-56-64
HySI ImageOrbit: 1089DOP: 07-02-09
Histogram1.5 km
Strip Width: 20 km
Chandrayaan-1 TMC IMAGE MOSAIC
Orbit: 3586 2580 3521
MOSAIC
DOP: 23-04 11-08 10-08StripWidth: 20 km 40 km 40 km
0 10 km
85 km
Jawahar Sthal Seen by TMC
TMC Nadir ImageOrbit: 2792DOP: 29-06-09
TMC Nadir Image Overlaid on Clementine Polar Image
TMC Image
Jawahar Sthal (MIP Impact Point:Lat.: - 89.76 Long.:- 39.40 )
Shackelton Crater
Shackelton Crater Scale
0 45 km
Scale
0 22 km
All Projections are in Polar Stereographic
Portion of SAR and MIS referenced image, with uncertainty circle (Diameter- 1km)
Air-borne HySI Registration1. HySI bands are inherently acquired in non-registered mode. Successive bands look at the same feature at a time interval
of 51.8 msec.
2. Air-borne platforms are subjected to a high attitude rate resulting in distortion in image features due to changes in scale and orientation.
3. The Correlation based operator is employed for visible bands and Mutual Information based operator for infra-red bands
4. Implemented multi-threaded registration software for speeding up computation of multi-CPU multi-core machines.
Band 256 ( Reference Band)Acquired Band 100 Registered Band 100
26
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
S. No Model accuracy (in pixels) (Fore)
Model accuracy (in pixels) (Aft) No. of
GCPs used
RMS Scan RMS Pixel RMS Scan
RMS Pixel
Case 1 2.5 2.3 2 2.5 32Case 2 2.5 1.0 1.3 1.0 36Case 3 1.2 1.3 1.1 1.1 94
S. No Along error(m)
Across error(m)
Radial error(m)
Case 1 3.6 9.6 10.3Case 2 4.4 8.0 9.1Case 3 7.3 6.3 9.6Case 4 5.5 6.5 8.5Case 5 3.8 12.8 13.6Case 6 13 10 16.4Case 7 2.7 3.7 4.6RMS(m) 5.9 8.1 10.3
Table 2 : SSTS model accuracy
Table 3 : Accuracy of ortho products for AFT camera
27
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Figure 1 Cartosat-1 Strip DEM (over 500 km) generated from SSTS approach
Recent Advances in Cartosat-1 Data Processing
28
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Figure 2 SSTS Performance analysis (Model Error)
Recent Advances in Cartosat-1 Data Processing
0
1
2
3
4
5
6
7
8
9
10
8-M
ay-0
5
8-J
un-0
5
8-J
ul-05
8-A
ug-0
5
8-S
ep-0
5
8-O
ct-
05
8-N
ov-0
5
8-D
ec-0
5
8-J
an-0
6
8-F
eb-0
6
8-M
ar-
06
8-A
pr-
06
8-M
ay-0
6
8-J
un-0
6
8-J
ul-06
8-A
ug-0
6
8-S
ep-0
6
8-O
ct-
06
8-N
ov-0
6
8-D
ec-0
6
8-J
an-0
7
Date of Pass
RM
S e
rro
rs
RMS erros scanRMS error pixel
29
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
CONCLUSIONS
The model accuracy achieved through SSTS approach is better than 10 m (circular error ~ 4 pixels at product, inclusive of all the processing errors further to SSTS). From the analysis on results of all available dates, SSTS performance is seen to be good in terms of realised accuracy for both along track and across track error. SSTS pre-adjustment results at GCPs were used for geometrical in-flight calibration analysis to estimate spacecraft misalignment to cameras, attitude biases and inter camera alignment angles. SSTS has been found to be successful in all cases when cloud free segments of stereo pair data and required number of GCPs (10 or more) were available for given pass.
30
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
Approach for restoration of Cartosat-1 imagery
The laboratory measured PSFs for the Fore and Aft sensors of Cartosat-1 were taken as the degradation function for restoration of respective imagery. Wiener filter, which incorporates the degradation function as well as the model of noise, was designed in frequency domain for each sensor. The noise to signal ratio was modelled to vary exponentially between nsrmin and nsrmax for low to high frequencies. Restoration of the FORE and AFT images was performed block-wise. Overlap of few pixels was maintained between successive blocks to avoid artefacts at block boundaries. Fast Fourier Transform (FFT) techniques were used to achieve high speed.
31
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
Area Date Cartosat-1 images(dd/mm/yy)
Path/Row Camera Location (lat./long. in Degs)
Rome 08/06/05 170/206 Fore 41.891/12.725Rome 08/06/05 170/206 Aft 41.897/12.705Brazil 15/09/06 1822/494 Fore -22.73/316.29Brazil 15/09/06 1822/494 Aft -22.73/316.29Japan 31/03/06 798/252 Fore 31.560/130.646Japan 31/03/06 798/252 Aft 31.560/130.642
Table-4 Data sets used for image restoration method
32
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
Figure 5 (c) ROME: Cartosat-1 AFT (Original)
33
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
Figure 5 (d) ROME: Cartosat-1 AFT (PSF corrected)
Japan512x512
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
Japan512x512
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
CARTOSAT-1INDIAN SPACE RESEARCH ORGANISATION
Recent Advances in Cartosat-1 Data Processing
Brazil640x640
Rome640x640
42
INDIAN SPACE RESEARCH ORGANISATION