This tutorial provides a one-day introduction to the theory and applications of morphological image processing. Mathematical morphology has proven itself as a powerful frame work for image analysis, particularly the analysis of shapes in images.
In this tutorial the following subjects will be discussed:- Morphological operators for sets (binary morphology) and their geometrical interpretation. The basic operators: erosion and dilation, are used to build operators to detect specific shapes (hit-or-miss transform), to remove shapes (thinning) or add them (thickening). A shape can be reduced to its 'skeleton' allowing for a more compact shape characterization. Furthermore we discuss distance transforms, granulometries (the morphological tool to assess the the size distribution of the shapes present in an image) and the important role of convex sets in morphological shape analysis.
- The mathematical foundation of all morphological operators is the 'complete lattice framework'. This framework is introduced and related to binary morphology. Within the framework we are then able to generalize the morphological toolbox to work with grey value images and even color images.
- Within the morphological framework a 'filter' is an image operator that is idempotent, i.e. after applying it once to an image, further applications of the filter will not change the image anymore. This is of course a desirable property when we are removing noise for instance. The morphological filter theory encompasses operators like rank-based operators and alternating sequential filters.
- The notion of 'connected operators' unifies all operators that deal with the connectivity of shapes visible in images. Reconstruction operators generalize the 'flood-fill' (propagation) operators from binary morphology. Other connected operators of practical value are: geodesical operators, area openings, grain operators and levelings.
- No tutorial on mathematical morphology would be complete without an introduction to the watershed segmentation algorithm. This is without a doubt _the_ standard segmentation method from the morphological toolbox.
- We hope that the tutorial provides you with enough background to appreciate the current trends in mathematical morphology: pyramids, wavelets, scale-space and color morphology. We can only sketch the horizon of these trends in this one day course.
The tutorial will be presented by Henk Heijmans and Rein van den Boomgaard. Both of them are more then 10 years involved in morphological research and teaching. This tutorial is based on a PhD-level course that they teach in the Netherlands every two years.
Participants are not assumed to have any prior knowledge on mathematical morphology. Prior exposure to the basic tools in image processing is helpful though.
The cost of this tutorial is A$ 400.