ENL computation on SAR image

I have level 1.5 image which I am using for speckle filtering. Somewhere it is written that speckle filtering is applied before calibration and somewhere it is specified to do filtering after calibration. I am not getting which one is correct? After filtering ENL is computed on sigma-naught image? Please help to clear my doubts. I am using snap 5.0.

Level 1.5 means that it is already multi-looked to a square pixel resolution. So the original values are alreay averaged some how. I would personally calibrate first because after filtering, you have clearly lost some information on backscatter (again). If you then later apply terrain correction, there is another resampling of the pixels so even without speckle filtering, the original values have been altered quite much.
The original speckle filters were designed to be applied on SLC data (Level 1.1), but if you are already working in ground range geometry, you could as well pre-process your data and see how much speckle is left after all required steps and then apply the filter as a final step.

Ok. Thanks for the reply.
Can you guide, how homogeneous regions are selected from image?

I would say that these are areas with the same surface characteristics. Depending on where you work, this can be agricultural fields, grassland, oben (calm) water bodies, larger paved areas. Anything that is ‘noisy’ within those helps to characterize the degree of speckle.

I got it. Thanks.