How can I use adaptive level of binarizing for different images in a loop?

I have thousands of images to find centroid. Here I show two images. So, the binarizing level needed for these two images are not the same and if I am using a loop for all my images, how to make some adaptive level for binarization of these?

Answer 1

[Image Analyst]

I don't see any signal in the second image - pretty much just noise. I'd suggest you just use a fixed threshold if you can get control over your exposure. That way it will detect something if it's there and detect nothing if nothing is there. If not, you may want to increase your exposure time to give your signal time to build up to a substantial level above the noise (increase the signal to noise ratio).