Multi-looking also increases pixel spacing, because you typically average a number of samples (e.g. 4 samples at 10 m spacing become 1 sample at 20 m). Speckle filtering does not change spatial detail.
I agree. I mostly use speckle filtering for image enhancement. But when I need squared pixels (from SLC products) or want to resample to a lower resolution I chose the multi-looking.
Multi-looking GRD products doesn’t make much sense to me just for speckle removal.
Thanks for good info.
If I want to integrate the SAR processed image with optical image, do I need to perform “multilooking” on sigma0 in order to have SAR squared pixels? (Radiometric Calibration (sigma0), “multilooking”,Terrain Correction)
In general, when it’s recommended most to get SAR pixels squared?
in all cases where you need the geographical location of the SAR information, for example when you want to use it in a GIS later or when collocating with other (optical) imagery.
But you will receive square pixels after terrain correction as well. So multi-looking before that step is not necessarily needed.
Following this discussion I noticed that when doing Iterferogram Formation processing there is also check box for square pixel. Could someone clarify this?
Thank you
most filters are designed for the original data and are less effective when applied to resampled or multi-looked data. Personally, I would reduce speckle first and then multi-look to get best results, but it is worth to try both and compare the outcomes to see which are more suitable for you.
well, filtering actively reduces speckle while the main purpose of multi looking is to reduce the spatial resolution and to harmonize azimuth and range resolution. As a side effect speckle is often also reduced, but not in an adaptive way as it is done by the speckle filters.