Zusammenfassung
An essential goal of image processing is the detection
of image structures like objects, object boundaries,
shapes, regions or surface defects. This goal can be
reached by a combination of detection algorithms and
localisation algorithms. Detection algorithms are able
to recognise the physical evidence of a structure
according to a given localization in the
image.
Localisation algorithms cope with the position of image
structures in images. Most of them are tracking
algorithms. Because of the limited recognition
abilities the main goal of the foregoing detection
algorithm is to enhance an image object. The image
structure must be transformed into a kind of
information which is suitable for the localisation
algorithm.
However, in many fields of image processing this
approach does not give satisfying results.
This paper introduces a new approach by replacing the
conventional localisation algorithm by an image
exploring agent and by adapting its functional design
by means of genetic programming.
Nutzer