On Thursday 30 Jan 2014 18:34:34 you wrote:
> I'm wondering what is the minimum object size (in pixel) for the box
> counting method using box sizes from x (standard value is 2) to y
> (standard value is 64) to find a good estimate of the object's fractal
> dimension.
> Is there a way to calculate such a minimal size?
No, unfortunately. But you can investigate in your data which is the range of
values that make the log-log plot more or less straight.
My suggestion is not to use too small boxes unless you are measuring a
"constructed" mathematical fractal because you are playing at the limit of
image resolution in captured images.
Too large boxes have a similar problem: you are just measuring the object
size.
Maybe this review in JMicrosc would be useful as it deals with several of
these technical issues.
http://www.ncbi.nlm.nih.gov/pubmed/21118245Cheers
Gabriel
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html
Locked
