There are several different watershedding approaches in the literature, and many applications of watershedding exist in binary and greyscale image processing, particularly in vision systems. There seems to be little work published which systematically compares and assesses the performances of these algorithms. This paper discusses the different watershedding approaches for greyscale images in terms of accuracy and speed of computation, and then compares the performance of three of the fastest watershed algorithms based on the flooding approach. Experimental results of the application of these algorithms to the segmentation of real images (teeth X-ray imaging) are presented and discussed.
[1]
Serge Beucher,et al.
Use of watersheds in contour detection
,
1979
.
[2]
S. Beucher,et al.
Watersheds of functions and picture segmentation
,
1982,
ICASSP.
[3]
Moncef Gabbouj,et al.
Fast watershed algorithms: analysis and extensions
,
1994,
Electronic Imaging.
[4]
Edward R. Dougherty,et al.
Mathematical Morphology in Image Processing
,
1992
.
[5]
Luc Vincent,et al.
Watersheds in Digital Spaces: An Efficient Algorithm Based on Immersion Simulations
,
1991,
IEEE Trans. Pattern Anal. Mach. Intell..