sarproz-group

S-1 coregistration error

This topic contains 8 replies, has 2 voices, and was last updated by  Yuxiao 10 months ago.

  • Author
    Posts
  • #1817

    espiritocz
    Participant

    Hi,

    using the latest (compiled) version, I got some error (attached) when trying to extract slaves.
    This was tried in two different datasets (same area, but ascending and descending tracks).
    I have manually checked the files and they should contain the overlapping area normally..

    Does the lasterror say something about corrupted images? (sometimes this can happen..)

    It is interesting that two different tracks show the same (similar?) error.

    Also – I have solved the problem in one of the tracks simply by removing the other SLCs (keeping dataset of 36 images, until March 2016) – but when I tried to coregister these images, there were 11 images not coregistered. I send also lasterror.mat files for those coregistration errors, hoping that (some of) you will know the cause of this problem…

    Thank you very much

    Milan

    Attachments:
    You must be logged in to view attached files.
  • #1820

    espiritocz
    Participant

    only adding a comment – the coregistration errors are accompanied by following message:
    Error in amp_corr(): Match parameters missing!

    do you have any clues?
    thank you

  • #1821

    Yuxiao
    Moderator

    Hi Milan,

    I will need the following things from you:

    1. the footprints of all your images;
    2. A printscreen of SLC import, including your master image, the center lat/lon, samples/lines;
    3. Try to see what are the slave images that are not extracted. You can do so by checking dataset.txt and see what is missing.

    I will need those info to check where’s the problem. Either attach them in forum or send me an email.

    Info from either problematic sites is fine.

    yuxiao

  • #1822

    espiritocz
    Participant

    Hi Yuxiao,

    thank you.
    So, I have realized that the coreg-erroneous slave images are cropped from totally different area!
    (it is strange, even the jpg previews were not generated, i had to do it manually and saw that they are after Slave Extraction from totally different location.. please note that I use precise orbits normally here…)

    Let’s try it only for one track (i choose track 51 where 27 images were coregistered well)
    I attach the footprints and screenshot.
    I know exactly which ones are not coregistered – there are 11 images:
    20151121
    20151203
    20151215
    20160108
    20160120
    20160201
    20160225
    20160320
    20160413
    20160425
    20160519

    I know which ones were not extracted. It seems that really some of them were corrupted. I am checking them one by one now..so, let’s say the problem is reduced only to this coregistration error, caused by wrong cropping of these 11 images.

    Thank you.

    Attachments:
    You must be logged in to view attached files.
    • #1824

      Yuxiao
      Moderator

      Hi Milan,

      I need you to do one experiment here:

      Create a new dataset, with ONLY the master image and all those problematic slave images.

      Then try to extract the master and slave image with the EXACT parameters you used for every step.

      Can you then check by “View Master” and “View Slave” if they roughly correspond to the same area?

      It should save you time by investigating into this even before you start doing coregistration.

      Then if you again see an error extracting slave images, please let me know.

      Yuxiao

  • #1825

    espiritocz
    Participant

    Thank you Yuxiao, in the meantime I tried something else:

    I have removed the corrupted files and removed also these wrongly cropped images, re-extracted slaves and – it is working normally now..

    I would like to recommend some test for corrupted files. Also, I don’t understand why the presence of one corrupted file caused that another file was cropped in different area..seems like this is what happened here.

    Anyway, it is indeed up to user to double check his files J

    So, thanks so far, I continue working. Cheers!

    • #1826

      Yuxiao
      Moderator

      We need to investigate if with a good SLC file it ended up extracting a wrong area. That is why I need you to keep these files and do some tests to see if indeed there’s a bug in extracting slave images in such a case.

  • #1827

    espiritocz
    Participant

    ok,

    so, I have tried this – i used the same master and put only 4 wrong slaves in a “test” project.
    Two of them were extracted (normally?):

    4 Slave images have not been extracted!!

    Extracting Slave Images

    SLC data extraction

    Your S1A imgfile: /media/TEMP_90DAYS/processing/test/SLC/S1A_IW_SLC__1SDV_20150825T163428_20150825T163455_007422_00A376_7C99.SAFE/measurement/s1a-iw2-slc-vv-20150825t163430-20150825t163455-007422-00a376-005.tiff seems to be corrupted and NO HEADER INFO can be read!!! Further Process with this image may lead to error!!!

    Your S1A imgfile: /media/TEMP_90DAYS/processing/test/SLC/S1A_IW_SLC__1SDV_20150825T163428_20150825T163455_007422_00A376_7C99.SAFE/measurement/s1a-iw2-slc-vv-20150825t163430-20150825t163455-007422-00a376-005.tiff seems to be corrupted and NO HEADER INFO can be read!!! Further Process with this image may lead to error!!!

    Your S1A imgfile: /media/TEMP_90DAYS/processing/test/SLC/S1A_IW_SLC__1SDV_20150825T163428_20150825T163455_007422_00A376_7C99.SAFE/measurement/s1a-iw2-slc-vv-20150825t163430-20150825t163455-007422-00a376-005.tiff seems to be corrupted and NO HEADER INFO can be read!!! Further Process with this image may lead to error!!!

    Your S1A imgfile: /media/TEMP_90DAYS/processing/test/SLC/S1A_IW_SLC__1SDV_20150918T163429_20150918T163455_007772_00ACE7_BE0B.SAFE/measurement/s1a-iw2-slc-vv-20150918t163430-20150918t163455-007772-00ace7-005.tiff seems to be corrupted and NO HEADER INFO can be read!!! Further Process with this image may lead to error!!!

    Your S1A imgfile: /media/TEMP_90DAYS/processing/test/SLC/S1A_IW_SLC__1SDV_20150918T163429_20150918T163455_007772_00ACE7_BE0B.SAFE/measurement/s1a-iw2-slc-vv-20150918t163430-20150918t163455-007772-00ace7-005.tiff seems to be corrupted and NO HEADER INFO can be read!!! Further Process with this image may lead to error!!!

    Your S1A imgfile: /media/TEMP_90DAYS/processing/test/SLC/S1A_IW_SLC__1SDV_20150918T163429_20150918T163455_007772_00ACE7_BE0B.SAFE/measurement/s1a-iw2-slc-vv-20150918t163430-20150918t163455-007772-00ace7-005.tiff seems to be corrupted and NO HEADER INFO can be read!!! Further Process with this image may lead to error!!!

    An error occurred and was saved in the file lasterror.mat, directory /home/insarist/SARPROZ

    If you need assistance, please include the file lasterror.mat

    You may include also file 23_Aug_2016_21_11_20_sarproz.log

    Here a glimpse:

    Too many input arguments.

    2 Slave images have not been extracted!!

    I tried to look at them, but by pressing “View Ex. Sl.”:

    File /media/TEMP_90DAYS/processing/test/EXT/20150825_VV.slc.jpg not found!!!

    File /media/TEMP_90DAYS/processing/test/EXT/20150918_VV.slc.jpg not found!!!

    File 20150930_VV.slc not found

    File 20151024_VV.slc not found

    So, I used my script to generate jpg from the extracted slcs (20150825 and 20150918) and checked them – but it seems bad, only a grey area. Probably really corrupted.

    Okay, I just created a script to check corrupted tiff files – it is possible to use gdal for it, for example.. and indeed, those files are corrupted.

    so…. what was the aim of this investigation?
    should i use also one good image?
    yes..

    okay, trying to include also one good image. again pressing SLC import (now there are 5 slaves – 4 corrupted and 1 good). the result is:
    the good one was extracted normally.
    So, I could not reconstruct this error – it seems it was caused by something else but I have no idea what caused it.

    So, I am sorry Yuxiao, I did not help in debugging….

    Thank you for your time.

    Milan

    • #1828

      Yuxiao
      Moderator

      It is good to know this is a corrupted file instead of a sarproz bug. Can you probably try it again when you download again the correct file and see if it works well?

      yuxiao

You must be logged in to reply to this topic.