fit polynomial & separating colors

[rtas]

Hello,

I have the histological serial images that taken at low magnification.
I would like to separate region of interests based on colors, like
cartilages in blue, neuclei in pink/red and mineralized bone in brown,
using color deconvolution.
Before this, I need to correct the uneven illumination of background.

What im doing for background corrections are as below:
1) use color threshold in RGB image, select the samples using brightness.
2) edit > selection >make inverse to select the background
3) plugins > fit polynomial [by Michael Schmid]


The brightness of the image after polynomial process results in darker
image overall and still the background seems to show the uneven
illumination.
Also the color of images have been changed, and its difficult to tell the
resulting color of images are correct.
Am I doing correction wrong?
Or is this because that the threshold in brightness of background is really
high and close to those of the objects?


Also is there any way to know the optimal x, y, and xy for fit polynomial?

I saw other polynomial plugins however, i thought fit polynomial plugins
would work for my sample since this plugin works with selecting the
background as region of interests for fitting.
I tried the subtract background by rolling ball, but the radius of ball
needs to be large enough not to select the regions of the sample, thus
it does not seem to work to subtract the background.

I really need to correct the background to process the color  segmentation
by color deconvolution.
I appreciate any suggestion!


Rui

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html