Selecting optimal parameters for cluster analysis applied to over-segmentation reduction in color images
Artykuł w czasopiśmie
| Status: | |
| Autorzy: | Smołka Jakub, Skublewska-Paszkowska Maria, Wojdyga Aleksander |
| Rok wydania: | 2012 |
| Wersja dokumentu: | Drukowana |
| Język: | angielski |
| Numer czasopisma: | 58 |
| Strony: | 115 - 127 |
| Efekt badań statutowych | NIE |
| Materiał konferencyjny: | NIE |
| Publikacja OA: | NIE |
| Abstrakty: | angielski |
| Watershed transformation is an image segmentation method whose main advantage is that it extracts almost all edges present in the image.Moreover all edges are always continuous,which is the algorithm`s main advantage.Unfortunately,the method frequently produces over-segmented images.The authors proposed a solution to the over-segmentation problem in color images.It utilizes hierarchical cluster analisys and treats watersheds as objects characteeized by a number of attributers.This paper presents research on selecting parameters for chuster analysis.The parameter sets that allow for obtaining the greates number of good segmentations of a given test image are selected.Conclusions regarding optimal parameters alongside with selected segmentations are presented. |
