Automatic Watershed Segmentation and Comparison between Pixel Scan Method and Order-Invariant Algorithm for Image Segmentation
Devappa Jayadevappa, S. Srinivas Kumar, D. S. Murty · 2007
The Watershed transform is a popular segmentation method coming from the field of mathe-matical morphology. It has interesting properties that make it useful for many different image segmentation applications. In this paper we present two approaches: pixel scan approach and order invariant based approach. In the pixel scan method unlike the traditional concept of use of chain codes, we here have used pixel-scan method, where all the pixels and their four neighbors are compared for various parameters assigned to them. The immersion approach can be viewed as an approach that starts from low altitude to high altitude. Here we present two order invariant algorithms for watershed segmentation. When running on a PC both our algorithms run fast for a 512 x 512 image, on average. What is more surprising is that the pixel scan based algorithm turns out to runs faster than the order invariant algorithm for almost all the images we have used. Especially when the image is large, say 512 x 512 or large. An experimental result shows that the presented algorithm runs fast. Further more the new algorithm can provide more information for image processing. Key words - Watershed transform, pixel scan, image segmentation, and order invariant.