Speckle filter (filters and window size)

Dear all,
I am wondering if anyone has experience on trying different filters and window size in Speckle filter in S1A images?

By default, it is leesigma, 7*7 and 0.9. Any analysis/comment?


1 Like

I asked a similar question recently - i was losing too much information when speckle filtering. It seems one answer is to try a multi-temporal speckle filter

Are any codes available in python for speckle filter. kindly guide.

Speckle mostly relates to the size of the scatterers relative to the resolution of the sensor. In “distributed targets” (e.g. a soil surface, crop canopy), the scatterers are much smaller than the resolution cells, and leading to “pure speckle”, which depends only on the (equivalent) number of looks. I would be surprised if co-pol and cross-pol would have different speckle statistics for those cases, unless the co-pol and cross-pol contribution would arise from very different scattering mechanisms.

The general idea of a speckle filter is to separate “pure speckle” from intensity variation that is due to non-speckle, e…g. when there are strong scattering targets (building, pylons, etc.) or when you are at the edge between 2 or more distributes targets with different backscattering intensities.

So, the choice of filter, window size and parameters settings depends on what you try to achieve. There is massive literature on speckle filters, all with some favourite choice and parameter settings, but often linked to some thematic application (agriculture, forestry, ocean, etc.).

1 Like