Multi Otsu Threshold
| Multi Otsu Threshold (ImageJ) | |
|---|---|
| Author | Yasunari Tosa |
| Maintainer | |
| File | Multi_OtsuThreshold.class |
| Source | Multi_OtsuThreshold.java |
| Initial release | 14 April 2006 |
| Development status | stable |
| Category | Segmentation |
| Website | [1] |
Purpose
This plugin segments the image in classes by thresholding. It uses the same algorithm found in Otsu Thresholding, but was adapted to output more than 2 classes out of the process.
Documentation
This plugin implements an algorithm described in the following paper Liao, P-S. & Chung, P-C. (September 2001), "A fast algorithm for multilevel thresholding", Journal of Information Science and Engineering 17 (5): 713-727, <http://www.iis.sinica.edu.tw/page/jise/2001/200109_01.pdf>
A thresholding algorithm will typically classify pixels in two classes (or two set of objects): the one that have their intensity lower than a certain threshold (generally, the background), and the other (the interesting features). This plugin is based on the Otsu Thresholding technique, adapted to generate multiple thresholds and multiple classes from one single image.
For example, by setting the desired number of classes to 3 (the algorithm then needs to find 2 thresholds), one can get background pixels, bright pixels and intermediate pixels. This might be of interest for images where there is such a pixel populations. In the example depicted below, based on the blob image, one could get the background, the blobs center and the blob edges out of it.
Original image: blobs on a dark background
First class: the dark pixels as background
Second class: the intermediate pixels delineate the edges
Third class: the bright pixels form the blobs centers
