Well, a way is to subtract the background with the paraboloid slide option. This should at least correct the uneven illumination. Even if you will have to play a bit the the parameters to see if your image gets better

There are various options: Gaussian blur and subtract, and a high pass fft filter are two examples. I'd be happy to take a look at your image and recommend something more specific. I also strongly encourage you take a look at our new software MIPAR at http://MIPAR.us. If offers a very intuitive interface for scientific image analysis and lets you build up non destructive recipes for preprocessing as well as feature finding if you are ultimately interested in that.

Cheers,

John

Hello Charchit,

If I understood your question correctly, I think you can eliminate/reduce random illumination and shading by carefully designing some lights. So this step in not within the image processing part, but you may find some image processing tools too. 

As an extra note for future, have you looked at the illumination setup itself?

Kohler illumination requires some alignment which is easy to do - it should sort out the problem at the hardware end unless something really odd is happening.

http://zeiss-campus.magnet.fsu.edu/articles/basics/kohler.html

Hey Charchit, have a look at this post from Kurt Thorn's blog: 

http://nic.ucsf.edu/blog/?p=1032

You will find some material on shading and uneven illumination correction.

Cheers,

Filippo

Hi Charchit,

A fast way to correct for uneven illumination could also be taking an image without your sample and then subtract i from the final image, with some scaling if needed. An image processing method I recommend is median subtraction that, compared with gaussian mean, is more conservative with sharp borders.

Cheers,

Lorenzo

Hi,

NIH ImageJ can substract background via rolling ball algorithm. You can also use FFT bandpass filter, or just substract an empty frame of illumination from your image. You can also use Virtualdub, remove bulb-effect filter.

Thanks you all for your informative suggestions and ideas.

@John: I will look over the software you mentioned. 

Please have a look of the sample image, on which I am doing my analysis. I am interested to measure the area of bright spot (circular bright spots in this image). The problem I am facing here is to remove some illumination coming from right bottom corner.

Hi Charchit,

Thanks very much for giving MIPAR a try. Based on your image, I think it will be very helpful. I will make up a recipe to find your spots and send it to you ASAP.

Cheers,

John

Hi Charcit,

Here is a BW and outline view of the segmentation I was able to achieve for the circular bright spots. Is this what you were after? Using MIPAR's adaptive threshold, followed by some of its other cleanup steps, I was able to avoid the need for shading correction of any kind. I have also attached the recipe file for you to load yourself. Let me know of any questions!

Cheers,

John