Otsu ThresholdingQR for this page
|Otsu Thresholding (ImageJ)|
|Author||Christopher Mei, Anthony Joshua, Tony Collins|
|File||Otsu_Thresholding.jar (15,549 Bytes)|
|Source||in .jar file|
|Initial release||15 December 2003|
|Latest version||15 June 2005|
This plugin automatically threshold an image according to Otsu's method, described in his paper:
- Otsu, N. (Jan. 1979), "A threshold selection method from gray level histograms", IEEE Transactions on Systems, Man and Cybernetics 9 (1): 62-66, CODEN: ISYMAW, ISSN 0018-9472
For each image of the stack, the plugin determines a threshold by histogram partitioning. The histogram is divided in two classes and the threshold is the value that minimizes the within-class variance. There is no use of spatial coherence, nor any other notion of object structure.
Assumptions of the algorithm:
- the histogram is bimodal.
- uniform illumination (implicitly), so the bimodal brightness behavior arises from object appearance differences only.
This plugin only operates on 8-bit images. A 16-bit version can also be found: OtsuThresholding_16Bit.java
It is a bit painful to use with stacks; it determines a threshold for each independent image, but does not output a stack (instead one get a collection of separate images) and bothers the user for each image with a result dialog at each step.