Multiscale methods which provide a decomposition of an image based on scale have many uses in image analysis. One class of such methods from mathematical morphology is based on size distributions or granulometries. In this paper a different type of image decomposition based on shape but not scale is proposed. Called a shape granulometry or shape distribution, it is built from a family of morphological thinnings, rather than openings as in the case of size distributions. An implementation based on scale invariant attribute thinnings is presented, and an example of an application is shown.
|Title of host publication||EPRINTS-BOOK-TITLE|
|Publisher||University of Groningen, Johann Bernoulli Institute for Mathematics and Computer Science|
|Number of pages||10|
|Publication status||Published - 2002|
- attribute thinnings
- connected filters
- multi-scale analysis