Window size in Polarimetric Decompoistion

Dear all,
I am new to PolSAR, I wonder when processing “Polarimetric Decomposition”, why do some algorithms (such as the H-Alpha Dual pol decomposition, which is the only method applicable to dual-pol S1) require entering a window size; while few others (such as Cameron’s and Pauli’s) do not require one?

as I would like to extract aquaculture features from the S1, yet after decomposition the resolution reduces significantly

Maybe can @jun_lu provide some references or explanation?