In [1] we present details of an image segmentation algorithm that was first introduced in my diploma thesis [2]. The segmentation is based on a height map which is generated in multiple steps from the gray values of the input image. Among the different methods for image segmentation morphological watersheds have some advantages. They yield more stable results in comparison to other segmentation concepts such as detection of discontinuities, thresholding, or region processing. But they also have a drawback. Watersheds work reliably only on height level images, i.e., images that are interpretable as some form of topography like images of cells under a microscope. In order to widen the applicability of watershed segmentation to arbitrary images such as photographs of natural objects we propose to generate a height map which characterizes the content of the image in an appropriate way. To do so we derive a height map from an edge filtered version of the input image enabling us to apply the watershed concept for the segmentation of arbitrary images. In addition, and in contrast to the general watershed concept, where markers are adopted to incorporate knowledge-based constraints in the segmentation process, we automatically generate markers from the height map to facilitate a more autonomous segmentation process.
The algorithm was featured on the cover of “Informatik Spektrum” [2], the main organ of the German Informatics Society (GI).
References
1
,
Image Segmentation Based on Height Maps,
In: Computer Analysis of Images and Patterns. Vol. 4673. Lecture Notes in Computer Science. Springer Berlin Heidelberg, pp. 612–619, 2007,
[pdf|doi|bibtex]
2
,
Dynamisches Lernen von Nachbarschaften zwischen Merkmalsgruppen zum Zwecke der Objekterkennung,
In: diploma thesis, University of Dortmund, 2006,
[pdf|bibtex]
3
,
Höhenbildbasierte Segmentierung,
In: Springer–Verlag, 2007,
[pdf|bibtex]
"Informatik Spektrum" cover