Create a Project by Logging In or Creating an Account

Multilevel Thresholding

Tue, 01/26/2010 - 04:04 pm

This is a first rough try at implementing "A fast multilevel thresholding algorithm" by Liao, Chen, and Chung, which is based on a "maximized between-class variance" criterion originally due to N. Otsu (IEEE Trans System Man Cybernetics, 1979).

In contrast to single threshold value obtained with the ImageThreshold operation, you can request up to four threshold levels to reduce the numbers of levels in a grayscale image. It has been said that MatLab has a similar operation called "graythresh" (see the forum topic initiated by Jason, who gave the idea for this project). More documentation about usage etc. can be found in the procedure file. Tested with IP6.10A on WinXP. Please report any problems via IgorExchange.

Images:

Project Details

Current Project Release