|
<< Click to Display Table of Contents >> Preferences specific - Adaptive Filter |
  
|
|
<< Click to Display Table of Contents >> Preferences specific - Adaptive Filter |
|
Purpose
This panel enables to set the default processing parameters, which are used in those functions related to the generation of interferometric products.
Technical Note
None.
Input Parameter(s)
This option enables to specify the Interferogram filtering method (i.e. Adaptive, Boxcar or Goldstein), which is used as default.
Coherence from Fint
By setting this flag the coherence image will be generated from the filtered interferogram (_fint); otherwise the unfiltered interferogram (_dint) will be adopted.
Feature Extraction Coherence from Fint
By setting this flag the coherence image will be generated from the filtered interferogram (_fint) during the feature extraction processing; otherwise the unfiltered interferogram (_dint) will be adopted.
Coherence Range Box Size (Goldstein and Boxcar)
Size (in range direction) of the moving window – in pixel odd number – for the coherence estimation. The default value can be increased working with lower resolution products (i.e. ScanSAR data).
Coherence Azimuth Box Size (Goldstein and Boxcar)
Size (in azimuth direction) of the moving window – in pixel odd number – for the coherence estimation. The default value can be increased working with lower resolution products (i.e. ScanSAR data).
Coherence Removal Local Frequency Range Box Size (Goldstein and Boxcar)
Size (in range direction) of the moving window – in pixel odd number – for the calculation of the local residual phase which is removed before the coherence estimation. The default value can be increased working with lower resolution products (i.e. ScanSAR data). It is mandatory that this value is higher than the "Coherence Range Window Size". By setting this value to 0, the residual phase removal is not performed for the coherence estimation.
Coherence Removal Local Frequency Azimuth Box Size (Goldstein and Boxcar)
Size (in azimuth direction) of the moving window – in pixel odd number – for the calculation of the local residual phase which is removed before the coherence estimation. The default value can be increased working with lower resolution products (i.e. ScanSAR data). It is mandatory that this value is higher than the "Coherence Azimuth Window Size". By setting this value to 0, the residual phase removal is not performed for the coherence estimation.
Goldstein Interferogram Window Size
Size of the moving window – better using power of 2 values – for the interferogram filtering. The bigger this window the lower is the filter sensitivity to small details (i.e. local fringe patterns). This value should be set between 32 (light filtering) and 512 (strong filtering).
Goldstein Window Overlap Percentage
A strong filter setting can introduce border line artifacts in the filtering window. These effects can be minimised/removed by increasing the window overlap percentage. The bigger this value the longer the processing time.
Goldstein Smooth Window Size
The power spectrum in convolved with a rectangular smoothing window, whose size can be specified in pixel odd number. The smaller this window the lower is the filter sensitivity to small details (i.e. local fringe patterns). This value should be set between 3 (strong filtering) and 11 (light filtering).
Goldstein Minimum/Maximum Alpha
It is the exponent applied to the power spectrum of the data. This is the most important parameter to tune the filter strength. In particular the "Alpha Min Value" is applied where the coherence is 1, while the "Alpha Max Value" is applied where the coherence is 0; in between them Alpha varies linearly from its minimum to its maximum value. The higher is Alpha (both Min and Max) the stronger is the filter smoothing.
The "Alpha Max Value" should vary between 0.5 (light filtering) and 4 (strong filtering); the "Alpha Min Value" should vary between 0.3 (light filtering) and 3 (strong filtering). When changing one of these two parameters, the other must be linearly modified. The paper suggestions refers to an alpha range from 0 to 1, SARscape can use values higher than 1 since they are remapped.
Goldstein Low Pass Percentage
This additional filter can be carried out in order to remove the phase noise (high frequency noise). This filter must be used carefully as it can remove also real phase information. The bigger the percentage the larger the frequency band, which is removed. If this value is set to 0 the low pass filter is not executed.
Boxcar Width
Size of the moving window – better using power of 2 values – for the interferogram adaptive filtering.
Boxcar SNR (Signal to Noise Ratio)
Minimum value of the Signal-to-Noise Ratio to carry out the interferogram filtering.
SNR = γ2/1-γ2
Adaptive Coherence Max Range Size
Maximum size (in range direction) of the moving window used for the interferogram filtering and coherence estimation.
Adaptive Coherence Max Azimuth Size
Minimum size (in range direction) of the moving window used for the interferogram filtering and coherence estimation.
Adaptive Mean Factor
Mean Intensity difference among the pixels within an area to be considered stationary. This value is used for optimising the interferogram filter.
Non-Local H factor
It modules the filter intensity. The higher the value, the higher the strength of the filter.
Non-Local T factor
It modules the filter intensity during the iterative processing. The lower the value, the higher details preserved.
Non-Local Average Window Size
The size of the weighted average performed by the filter. By increasing the window size, the waving effect over homogeneous areas is reduced. However, the processing time exponentially increases in function of the window size.
Non-Local Similarity Kernel Size
The kernel size used to quantify the similarity probability (weight) for the weighted average.
Non-Local Similarity Min Kernel Size
This parameter activates the similarity probability kernel adaptivity (if this value is lower than the similarity kernel size).
Non-Local Number of iteration
Number of iterations. It is mandatory an odd number, where 1 means that no iteration is applied. For values higher than 3, it provides minor improvements with higher processing time.
Non-Local Rem Local Freq FTT Size
Local frequency estimation is done via FTT over the moving window of size "Non-Local Average Window Size". This parameter must be larger than "Non-Local Average Window Size" and must be a power of 2 (32, 64). This parameter is the padding for the sub-pixel local frequency estimation. The local frequency estimation is used to flatten the input interferogram before the similarity estimation.
Non-Local Rem Local Freq SNR Threshold
Minimum Signal to Noise Ratio to consider valid the local frequency estimation. When it is rejected, the software considers it as 0 (no flattening before the similarity estimation is performed).
Non-Local Boxcar Window Size
Size for the pre-processing coherence Boxcar estimation used as initial factor of the central pixel modulation.
Non-Local Aggregation Window
Size for post-processing aggregation (extra-smoothed result factor).
Non-Local Min Looks Number
Minimum expected number of looks (number of aggregated pixels for the similarity weighted means). In case of looks lower than the "Min looks number", the most similar pixel weights are clipped.
Non-Local CPM Alpha
Central Pixel Modulation Alpha factor, where Alpha is exponent for the pixel coherence difference in the central pixel similarity weight modulation.
Non-Local CPM D0
Central Pixel Modulation D0 factor, where D0 is the minimum coherence difference for the central pixel similarity weight modulation activation.
Non-Local Lower Bound Threshold (dB)
Threshold on calibrated power values to reject dark objects outliers from the similarity weighted mean estimation.
Values lower than this threshold (i.e. dark points identified as water) are not considered in the neighborhood filtering selection.
Non-Local Upper Bound Threshold (dB)
Threshold on calibrated power values to reject bright outliers objects from the similarity weighted mean estimation.
Values higher than this threshold (i.e. bright points identified as point target) are not considered in the neighborhood filtering selection.
Non-Local Ratio Threshold (dB)
Threshold on ratio between calibrated power values to reject the outliers from the similarity mean estimation.
Ratio values between the observed and the neighborhood point higher than this threshold are not considered in the filtering process. This parameter is associated to the Non-Local Lower Bound Threshold (dB) or Non-Local Upper Bound Threshold (dB) in neighborhood filtering selection.
Non-Local Equivalent Number of Looks
The Equivalent Number of Looks is equivalent to the number of independent Intensity values averaged per pixel during the multi-looking process. This parameter can be easily estimated over a homogeneous (stationary) sample in the input Intensity data according to:
ENL = mean2 / standard deviation2
If the ENL is not set (i.e., it is set to the special value -1), it is automatically estimated via an analysis over the whole image.
The higher ENL value, the lower the filtering power (H factor and T factor are reduced) in order to avoid over-smoothing.
General Functions
Load Preferences
It allows loading specific Preferences tthirteen different settings as possible alternative default processing values.
Load
It allows loading an .xml, .sml or .txt file where SARscape common Preferences where previously saved.
Save
It allows saving SARscape common Preferences for a future processing using the same preferences characteristics.
Help
Specific help document section.
Ok
The selected default processing parameters are loaded.
Cancel
The window will be closed.
Specific Function(s)
None.
References
None.