Avtor/Urednik		Lezoray, Olivier
Naslov		Supervised automatic histogram clustering and watershed segmentation. Application to microscopic medical color images
Tip		članek
Vir		Image Anal Stereol
Vol. in št.		Letnik 22, št. 2
Leto izdaje		2003
Obseg		str. 113-20
Jezik		eng
Abstrakt		In this paper, an approach to the segmentation of microscopic color images is addressed, and applied to medical images. The approach combines a clustering method and a region growing method. Each color plane is segmented independently relying on a watershed based clustering of the plane histogram. The marginal segmentation maps intersect in a label concordance map. The latter map is simplified based on the assumption that the color planes are correlated. This produces a simplified label concordance map containing labeled and unlabeled pixels. The formers are used as an image of seeds for a color watershed. This fast and robust segmentation scheme is applied to several types of medical images.
Deskriptorji		IMAGE PROCESSING, COMPUTER-ASSISTED COLOR