How to estimate a grayscale image background ?

Hello there, first question in this forum

I have to analyze automatically a bunch of grayscale images, including the intensity of some zones. I have those zones stored as bw2label's numbers, but I need to substract the background intensity from the medium intensity I have calculated on those areas (the sum of the pixel's intensity divided by the area)

These are fluorescence images, so all the images have a little glow, some little, some more. I was wondering if there is a function in which, you input the image, and it returns the estimated intensity of the background (it would be the medium value of all the points that arent included in the areas (stored as bw2label's numbers)). By the way, those areas are way more intense than the background, so if the function doesn't need to be input the areas to exclude, its not a problem.

Oh, and I am using Image Processing toolbox

Thank you very much for your time, and sorry for my english.

[Image Analyst]

You forgot to attach the image!

Answer 1

[Image Analyst]

You might try a two pass algorithm. The first pass you estimate the blobs. This is what you're doing now. Then you look at what remains. If your background is slowly enough varying that it can be modeled by a 2D polynomial, you take those pixels and put them into polyfitn() by John D'Errico. Then you subtract the estimated background model from the image and now you can do the second/final pass on the background-corrected image. Attached is a demo.