Correcting baseline with topHat

There are moments when an image doesn't have a perfect lightning and you have uneven background like here.

So, when extracting regions of interest the regular way you will have a mask like that:

//The default algorithm used is
//"otsu".
mask = image.threshold();

The cause of this is the nature of the threshold algorithm. In ImageJS it takes a global threshold variable, which means that the algorithms calculate one value and compare it with all the pixels to create a Mask. Obviously, such approach is not very effective when there are smooth shades and uneven illumination. One of the ways to deal with uneven illumination and improve the output is to apply topHat() method. It is a method that subtracts the result of morphological opening from the original image.

image = image.grey().invert();
//Big structuring element is necessary for
const kernel = Array(101).fill(Array(101).fill(1));
image = image.topHat({ kernel });

Therefore if we look at the histograms of these two images at a given point we will see that the intensities are spread more evenly after topHat, rather than before.

Histogram before applying topHat

Histogram after applying topHat