Watershed based region growing algorithm

Jakub Smółka


This paper presents a solution to a major drawback of watershed transformation: over segmentation. The solution utilizes one of its main advantages - very good edge extraction. It is a method that simulates pouring water onto a landscape created on a basis of a digital image. Unfortunately transformation produces a region for each local minimum so, usually, the number of watersheds (catchment basins) is too big. Watershed region growing is based on a minimum variance region growing algorithm [1-3]. It differs from the original in that it grows a homogenous region by adding and removing entire watersheds (catchment basins) and not separate pixels. The generalized watershed based region dilation and contraction are presented. Thanks to the use of watershed transformation, the region growing process is not able to grow a region easily outside the object boundaries. Test segmentations of two class images and a comparison between the minimum variance and the watershed based region growing are presented.

Full Text:


DOI: http://dx.doi.org/10.17951/ai.2005.3.1.169-178
Data publikacji: 2015-01-04 00:00:00
Data złożenia artykułu: 2016-04-27 10:14:21


  • There are currently no refbacks.

Copyright (c) 2015 Annales UMCS Sectio AI Informatica

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.