There are many algorithms for removing the effect of non-uniform background from images. If the non-uniformity is additive, it is sometimes useful to fit a polynomial to various points which you associate with the background and then subtract the resulting polynomial surface from the whole image. If the non-uniformity is multiplicative, you need to generate an image corresponding to the polynomial surface and use it to scale the original image.

The following is an example of removing additive background. We created the image by adding thea gradient 0.01*x*y to the original blob image.

In order to use the operation ImageRemoveBackground, we need an ROI mask designating regions in the image that represent the background. You can create one using one of the ROI construction methods.

Blobs image with gradient added signal. Red pixels indicate selected ROI representing the background.

Blobs image after removing additive background.