From the Image to the Map Geometrical Preprocessing for Security

1
From the Image to the MapGeometrical
Preprocessing for Security

• Karlheinz Gutjahr
• Joanneum Research
• Institute of Digital Image Processing

2
Motivation

• Requirements in a crisis situation
• Fast mapping
• Comparability with available data sources
• Objectives
• Support of various remote sensing (imaging)
  sensors
• Detailed knowledge of error sources

3
Image analysis workflow
Geometrical pre-processing
4
Image distortions
Perspective, optical line scanners
SAR
5
Rectification methods

• Based on non linear parametric sensor model
• Co-linearity, Range/Doppler,...
• Two directions
• Backward / indirect / map-to-image
• Forward / direct / image-to-map

6
Backward rectification

• Map-to-image transformation
• State-of-the-art method
• Fast algorithm
• Rectification artefacts
• Algorithm
• E,N of output frame
• H of DEM/ellipsoid/
• Line-of-sight
• Intersect with image plane

7
Forward rectification

• Image-to-map transformation
• Algorithmic background simple
• Height start value and increment

• Algorithm
• Line-of-sight
• H Start height
• Intersect with surface gt E, N
• ? H DEM value at E, N
• If ? ok end else H H incr. gt 3.

8
Error impacts
Displacement due to height errors
Displacement due to errors of sensor model
(pointing errors)
9
Sensor development

• with respect to pointing accuracy (circular
  error at confidence level of 90)

10
A-priori scenarios

• Position and orientation unknown (e.g. Landsat)
• State vectors (orbit positions) for defined
  acquisition time given (e.g. every second)
  different level of accuracy (Spot ....)
• Orientation (off-nadir incidence angle, along
  track tilt angle) approximately known (Spot,
  JERS, Aster, .....)
• Precise measurement of position and orientation
  parameters (Spot 5, ERS, Envisat, airborne
  optical and SAR scanners)
• Verification of accuracy
• Optimisation of orientation elements if
  insufficient

11
Basis (Ground) Control points
12
Least Squares parameter adjustment

• Per control point 2 equations
• Optical (modified) co-linearity equations
• SAR azimuth and range equation
• Rational polynomial coefficients
• Measured image coordinates (x, y)
• Known ground coordinates (X, Y, Z)
• Approximately known elements of exterior and (if
  applicable) inner orientation (?unknowns)
• Objective optimisation of elements of exterior
  and (if applicable) inner orientation

13
Error impacts
Displacement due to height errors
Displacement due to errors of sensor model
(pointing errors)
14
Sensor development

• with respect to nominal pixel resolution
  (panchromatic channel at nadir)

15
Topographic effects vs. pixel resolution
16
Reference (SRTM) DTM

• Resolution
• Horizontal 25 m (GLOBE 1 km)
• Vertical 1 m
• Accuracy
• Horizontal 20 m
• Vertical 4 m (1?)
• Tiles
• 15 x 15

SRTM
GLOBE
Source http//www.dfd.dlr.de/srtm
17
Height definition
ellipsoidal, orthometric, dynamic,
18
Example Quickbird multispectral
Rectified using SRTM C-band DEM and sensor
geometry
19
Example Quickbird multispectral
Additional image-to-image co-registration
20
Summary

• Motivation of geometric pre-processing
• Fast and comparable mapping
• Basis of image analysis workflow
• Rectification methods
• Rigorous / polynomial / interpolative
• Forward / backward
• Error impacts
• Topography / sensor geometry
• Sensor development

21
Conclusions

• Theoretical background of geometric
  pre-processing is known
• Limited number of generic sensor geometries
• Limited number of generic rectification methods
• Interpretation is different
• Not applicable for security scenarios
• New topics are arising (height definition)
• Optimization of some pre-processing steps still
  required (stereo DSM extraction)
• Comparability of results has to be ensured

22
What GMOSS can offer

• Definition of geometric pre-processing methods
• Common understanding of geometric pre-processing
• Benchmarking of methods used within NoE
• Alternatives to conventional techniques
• Best practice rules
• Definition of quality parameters
• Definition of data / metadata exchange format(s)