You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Fusion of sensor data, knowledge sources and algorithms for extraction and classification of topographic objects
Baltsavias, Emmanuel P.

International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 7-4-3 W6, Valladolid, Spain, 3-4 June, 1999
GCPs and the method of measuring their pixel coordinates. The
localisation accuracy usually lies in the 0.5-1 pixel range.
3.2.2. Geocoding
Through geocoding, image data are transformed to a common
reference map projection, e.g. defined through topographic
maps being used. This is the standard procedure for remote
sensing images. Displacement errors caused by topographic
relief are removed through the integration of a DEM (in such a
case, also the terms orthorectification and differential
orthorectification are used).
2. As the above transformation yields an irregular raster of
points, a proper grey value interpolation/resampling has to
be applied in order to get a regular raster.
Grey value
Figure 5 shows a general scheme of the image geocoding
procedure. Based on a DEM and an optimised imaging model
(see above), geocoding can be performed in two basic
processing steps:
1. First, for each output pixel which defines a co-ordinate
triple (East, North, Height) in the output map projection, the
corresponding location in the input image is determined
through a so-called map-to-image transformation.
2. Subsequently, a grey value is interpolated/resampled from
neighbouring input image pixels and assigned to the related
output pixel.
3.2.3. Parametric Image-to-lmage Registration
Alternatively to geocoding, image data can be registered in
image geometry. Based on parametric methods this can be done
through a so-called map-to-image transformation of geocoded
image data into the geometry of a reference image. In the
approach implemented at the Institute of Digital Image
Processing, the images to be registered first have to be
geocoded using the standard geocoding method. The
transformation of the geocoded images to the geometry of
another selected reference image is then done in a direct
resampling approach in two steps (see Figure 6):
1. For each geocoded input pixel the corresponding output
pixel location, i.e. column and line values related to the
reference image, is calculated using the transformation from
map coordinates to the reference image.
Fig. 5. Digital image geocoding workflow.
Fig. 6. Workflow of transformation from map to image