Abstract:
In the case of every quantitative image analysis system a very important
issue is to improve the quality of images to be analyzed, in other words, their
pre-processing. As a result of pre-processing, the significant part of the
redundant information and disturbances (which could originate from imperfect
vision system components) should be removed from the image.
Another particularly important problem to be solved is the right choice of
image segmentation procedures. Segmentation essence is to divide an image
into disjoint subsets that meet certain criteria for homogeneity (e.g. color,
brightness or texture). The result of segmentation should allow the most
precise determination of geometrical features of objects present in a scene
with a minimum of computing effort. The measurement of geometric
properties of objects present in the scene is the subject of image analysis.
