Hierachical cluster analysis methods applied to image segmentation by watershed merging
Artykuł w czasopiśmie
| Status: | |
| Autorzy: | Smołka Jakub |
| Rok wydania: | 2007 |
| Wersja dokumentu: | Drukowana | Elektroniczna |
| Język: | angielski |
| Numer czasopisma: | 1 |
| Wolumen/Tom: | 6 |
| Strony: | 73 - 84 |
| Efekt badań statutowych | NIE |
| Materiał konferencyjny: | NIE |
| Publikacja OA: | NIE |
| Abstrakty: | angielski |
| A drawback of watershed transformation is over-segmentation. It consists in creating more classes than there are objects present in the image. Over-segmentation partially results from the fact that the transformation extracts almost all edges present in the image, even those which are very weak. To alleviate this problem images are preprocessed: blurring (or selectively blurring) filter is applied before the edge detection performed by a gradient filter. Additionally, the resulting image may be thresholded in order to eliminate small gradient values. This paper presents an alternative solution to this problem. The solution uses the hierarchical cluster analysis methods for joining similar classes of the over-segmented image into a given number of clusters. First, it calculates attribute values for each class. Second optionally, the values are standardized. Third, cluster analysis is performed. The resulting similarity hierarchy allows for simple selection of the number of clusters in the final segmentation. Several clustering methods, including the Complete Linkage and Ward's method along with many similarity/dissimilarity measures have been tested. The selected results are presented. |
