Open-source TomoSAR package for PSDSInSAR and ComSAR algorithms

Dear colleagues,

We are pleased and honored to announce the first open-source TomoSAR
package to jointly handle PS and DS targets
(GitHub - DinhHoTongMinh/TomoSAR).

It offers the PSDS InSAR and novel ComSAR algorithms. To make it more
practical, we exploit the SNAP as an InSAR processor and StaMPS as an
InSAR time series tool. These tools are open source so that people can
apply our PSDS and ComSAR methods for an end-to-end processing chain.

More insight about the algorithms can be found here:
Ho Tong Minh, D.; Ngo, Y.-N. Compressed SAR Interferometry in the Big Data Era. Remote
Sens. 2022, 14, 390. Remote Sensing | Free Full-Text | Compressed SAR Interferometry in the Big Data Era

Thank you for your consideration.

Best wishes,
Dinh Ho Tong Minh and Yen Nhi Ngo

6 Likes

Thank you - it’s great to see a brand-new InSAR method released directly into the public domain!

1 Like

thanks for the package. i will use this for my work

Dear; @htmdinh

İ tried the scripts my study area and I found some unexpected result about point density.

I am working with a small area. I tried 3 different methods like your publiction and my result doesnt seems like yours.

First of all I started PSIinsar with snap2stamps python code an I produce 15464 ref PS.

Deramping computed on the fly.

**** z = ax + by+ c

15464 ref PS selected

Color Range: -7.75158 to 6.7679 mm/yr

After that I tried ComSAR with your scripts and catch 15470 ref PS.

Deramping computed on the fly.

**** z = ax + by+ c

15470 ref PS selected

Color Range: -7.76548 to 6.6264 mm/yr

And finally I tried PSDS with your scripts and catch 19101 ref PS.

**** z = ax + by+ c

19101 ref PS selected

Color Range: -8.86418 to 6.64183 mm/yr

I am expected much more PS points with comsar but the result very close to PSI results (maybe the same) The psds results create more points than comsar and PSI but I dont besure is it expected or not.

Do you have any idea why my results doesnt have enough PS points and not to look like your publiction results.

PS: İ used Sentinel-1 63 image asc and snap2stamps-masterv python code.

I dont change my parameter_input settings.

iance matrix.

CalWin = [7 25]; % - [row col]

Alpha = 0.05;

BroNumthre = 20;

Cohthre = 0.25;

Cohthre_slc_filt = 0.05;

ComSAR_flag = true;

miniStackSize = 5; %

Unified_flag = true;

Thank you…

Hi @suat,
Thanks for your question.
Can you have a look at the coherence of phase linking and the number of SHP? It looks like there is a small room for DS targets.
Since your dataset is quite small, can you send it to me to better check what is going on?
Kind regards,
Dinh
PS: Please post this kind of query in this group: Radar Interferometry community

Hi again @suat,

After checking your dataset, I suggest to use 0.4 as the thresholding. With ‘weed_standard_dev 0.8’, you should get something like 1020 points for PSI and 9707 for ComSAR.
Cheers,
Dinh

1 Like

Hi @htmdinh

Finally i am listening your advice and try the scripts with 0,4 threshold;

These are results;

COMSAR : 9711 PS points

İNSAR: 1129 PS points

PSDS: 13933 PS points

İts works you are right but i have one more question.Do you have any idea Why my PSDS results generate much more points then COMSAR results?

Thank you…

1 Like

@suat,

It looks nice now.

With full time series ComSAR version (Unified_flag = true), it is reasonable.

With compressed version (Unified_flag = false), you will get 20416 points like the attachment
ComSAR_comp

Cheers,
Dinh

1 Like

Dear @htmdinh

Which parameters are you using in Parameter_input scripts and Stamps? When i tried to changed (Unified_flag = false) and run mt_prerp_snap_comsar script i took 3157 points with stamps.

select_method: ‘PERCENT’

Dear @htmdinh

Thank you for all explanation. Maybe you can make a webinar for INSAR users about ComSAR. İ think this scripts very useful and deserve better understanding.

See you again…

You are welcome @suat.
I plan to give a tutorial for that at the IGARSS 2022.
IEEE IGARSS 2022 || Kuala Lumpur, Malaysia || 17 - 22 July, 2022
Welcome to join it!

Dinh

2 Likes