ADAPTIVE FEATURES OF DIGITAL ORTHORECTIFICATION

11th International Multidisciplinary Scientific GeoConference SGEM2011, www.sgem.org, SGEM2011 Conference Proceedings/ ISSN 1314-2704, June 20-25, 2011, Vol. 2, 313-320 pp

ABSTRACT

Orthorectification is main and specific photogrammetric task. Digital technologies allow generation of precise orthoimages, which are obtaining the pixel accuracy. The realization of this process requires usage of precise and adequate object model that is
generated by usage of different data sources. They include irregular set of points created
by survey measurements, digitized cartographic data, raster DEM obtained from stereo pairs, or from points’ cloud gathered by laser scanning.

The main steps of orthorectification are discussed that include georeferencing,
orthorectification and mosаicing of images. Georeferencing ensures connection between
image coordinates of points and corresponding co-ordinates of georeferenced control
points. This allows calculating the parameters for geometric transformation depending
on the used model. The preliminary step is generation of digital elevation model. The
complex model contains the vector data of break lines in object and data sets for every
one facets of the model. Automatic DEM extraction from stereo pairs requires the
definition of properties of areas for which the model is generated. The orthorectification
step is controlled by the transformation parameters, the used object model and is taking
in consideration the hidden parts of the object. The quality of rectified image depends
also on the method for grey level interpolation of adjacent pixels. The mosaic forming
requires correct segmentation of images and adequate combination the subparts of the
images. Important meaning has the method of definition the cutting line between adjacent images, radiometric corrections of mosaic parts and pixel level interpolation in the strip of connection.

The control of orthorectification process could be realized by interactive definition of methods and parameters of georeferencing, rectification and mosaicing. The selection of parameters could be done automatically on the basis of formulation the object type, the
definition of model restrictions and taking into account the image properties obtained by
image analyses. The artificial intelligence principles could be applied for
implementation the generation of object model and the orthorectification process.

Keywords: Digital Orthorectification, Image Understanding, DEM, Building Modeling, Artificial Intelligence