|
<< Click to Display Table of Contents >> General Tools - Data and Quality Analysis - RFI Filter |
  
|
|
<< Click to Display Table of Contents >> General Tools - Data and Quality Analysis - RFI Filter |
|
Purpose
This tool removes RFI interferences from SLC data in the frequency domain improving the image quality without degrading the image resolution and revelling terrain features that were previously covered by the RFI. As a second product dumped intermediate files can help to understand the nature of the interference in the area.
These effects affect SAR images since they are are susceptible to Radio Frequency Interference (RFI). Radio Frequencies are external signals that are received by the sensor during acquisition which can cause severe degradation of the image. They are usually produced by communication transmissions, air traffic or meteorological radars and military equipment whose operating frequency overlaps with the frequency band of a SAR system. The effects can be typically identified on SAR images as bright spots, lines or a general systematic noise over the image.
Technical Note
RFI signals that compromise the image quality must have a high interference-to-SAR signal ratio. This implies that RFI usually appear as various kinds of bright stripes (vertical for tones, horizontal por wide band pulsed signals) in the azimuth-time range frequency diagram of a SAR image. The usual technique to remove them is to notch those frequencies when present.
As this tool ingests SLC images (after range and azimuth compression), azimuth-time range frequency diagram cannot be straight forward calculated as the image is already focused. Hence, before the RFI removal the image must be defocused in azimuth.
Once defocused, the azimuth-time range frequency diagram is calculated and the RFI frequencies can be detected. In order to differentiate these signals from the actual SAR echoes, the software compares the power of a given sample with the average power of its neighbors. If this power is over a threshold it is marked as interference. Subsequently, this sample is notched by setting it as zero.
As this classification can be noisy, a second step is performed to refine the filter. As RFI signals are usually pulsed wide band signals or continuous tones, we expect that the interference pattern is not isolated as single pixels in the azimuth-time range frequency diagram. To remove those isolated samples from the filter an open morphological filter is applied and subsequently reconstructed by a close morphological filter.
Note that applying this filter to a non-contaminated image may degrade it. This tool performs better on StripMap images than on other modes (SPOT, TOPS, ScanSAR).
Input Files
Input file list
Input file name(s) of the Single Look Complex (_slc). Intensity data are not allowed. This file list is mandatory.
Parameters - Principal Parameters
RFI threshold
Threshold in dB to differentiate RFI signals from SAR echoes. Samples that are stronger than the threshold compared to the surrounding average are marked as RFI.
Smooth win
Number of neighbor pixels to average to compare with a given sample for detection.
Open az kernel size
Number of azimuth pixels of the open filter kernel.
Open rg kernel size
Number of range pixels of the open filter kernel.
Close az kernel size
Number of azimuth pixels of the close filter kernel.
Close rg kernel size
Number of range pixels of the close filter kernel
Tip: To find the correct set of parameters:
1.1.Turn Delete Temporary Files (in Global parameters) to False to dump intermediate files.
2.Set Open and Close Kernel sizes to -1. This will deactivate open and close morphologic filtes.
3.With the *_filter and *_filtered files as reference modify RFI threshold and Smooth win parameters until you are able to cover most of the RFI signal minimizing the notching of signal out of the RFI areas.
4.Once pleased with the initial filter start removing the small noisy spots by activating the open filter (i.e. starting with one pass and a 1 azimuth by 10 range kernel*). Effects of this step can be appreciated on _freq_filter or freq_filtered files.
5.Once a good removal of the noisy spots is achieved without removing the RFI areas proceed to enlarge these remaining areas with the close filter (ie. start with one pass and a 3 azimuth by 20 range kernel*). Results of this step are displayed on _freq_filter or freq_filtered files.
6.Verify the results looking at the _slc file.
Note: If data is affected by tones (vertical line on the az_vs_freq file) invert the morphologic kernels sizes.
Parameters - Global
It brings to the general section of the Preferences parameters. Any modified value will be used and stored for further processing sessions.
Parameters - Other Parameters
It brings to the general section of the Preferences parameters. Any modified value will be used and stored for further processing sessions.
Output Files
Output file list
Output file name(s) of the filtered data. This file list is mandatory.
_az_vs_freq
Azimuth time vs range frequency diagram (after azimuth defocusing).
_az_vs_freq_filter
Initial azimuth time vs range frequency RFI filter before the morphologic filters.
_az_vs_freq_filtered
Filtered azimuth time vs range frequency diagram (after azimuth defocusing).
fil_slc
Filtered SLC image.
Details specific to the Units of Measure and Nomenclature of the output products can be found in the Data Format section.
General Functions
Exec
The processing step is executed.
Store Batch
The processing step is stored in the batch list. The Batch Browser button allows to load the batch processing list.
Close
The window will be closed.
Help
Specific help document section.
Specific Function(s)
None.
Task, SARscapeBatch object, SARscapeBatch script example
References Reigber, A. and L. Ferro-Famil, “Interference suppression in synthesized SAR images,” IEEE Geosci. Remote Sens. Lett., vol. 2, no. 1, pp. 45–49, Jan. 2005.