sarproz-group

Sparse points selection

This topic contains 10 replies, has 2 voices, and was last updated by  periz 2 years, 4 months ago.

  • Author
    Posts
  • #972

    seppi-santiago
    Participant

    Hello Daniele,

    I am having a problem with Sparse Point Selection (both in the APS Processing and/or the Sparse Points Selection section which is in blue) . The process of selection seems to get stucked at the Mask Loading step. In previous opportunities I made tests with smaller datasets, of only 3 images, but now I’am working with a dataset of 9 images. I don’t know whether the number of images in the dataset should affect the Load of the Mask or not. If that is the case maybe I should just wait for it to finish.
    Just in case I enclose the Mask file (maschera.mat), there is no log file written for the session yet.

    Thanks,
    Santiago

  • #973

    seppi-santiago
    Participant

    Daniele,

    I’m sorry, the mask zipped file exceeds the maximum allowed limit

  • #979

    periz
    Keymaster

    the mask file is independent from the number of images
    it would be nice to know how big is maschera.mat, and which is the size of your dataset (samples and lines).
    by the way, which option did you use for mask generation?
    are you working with windows? what’s the size of the RAM?

  • #985

    periz
    Keymaster

    there is no evident reason for your problem
    the file is 2.2MB, less than 1.5e6 points
    loading the file in matlab takes a few seconds
    I am wondering whether you could have problems affecting matlab and/or windows installation.
    or maybe some problems in reading/writing operations: is the file on a local or external disk?
    which kind of disk are you using? is windows having enough free space in the installation disk?
    other question: which parameter are you trying to read?
    The sw firstly loads the mask, then it uses the sparse points to read your parameter (if it’s a full matrix, this is the operation that can take a bit of time more. in particular if you are using a slow external disk).
    if you press CTRL+C at the matlab prompt when it gets stuck, you could show me the error message and we may get more hints

    • #987

      seppi-santiago
      Participant

      Hello,

      The file is on a net disk, so I think this should not cause major problems. Windows has more than 500 available gb, and the usage of RAM memory is about 50% when running this command. The parameter from which I’m trying to select the points is the Amplitude Stability Index (Amp. Stab. Index 1-Sigma/Mu). I enclose a screen shot, since in my compiled windows version CTRLC + C causes the termination of the program.

      Thanks,
      Santiago

      Attachments:
      You must be logged in to view attached files.
  • #989

    periz
    Keymaster

    On the contrary, I believe the net disk is the major problem.
    Can you check the network usage?
    Or can you simply try to copy the dataset locally? Note that you do not need the SLC folder. So, I believe the rest of the folders should be quite small.

  • #990

    periz
    Keymaster

    On the contrary, I believe the net disk is the major problem.
    Can you check the network usage?
    Or can you simply try to copy the dataset locally? Note that you do not need the SLC folder. So, I believe the rest of the folders should be quite small.

    • #992

      seppi-santiago
      Participant

      Indeed, you were right. Now that I’m working locally the mask loading works perfectly. Thanks!

  • #1004

    seppi-santiago
    Participant

    Hello d,

    Normally I’ve created the mask with the ‘Local Maxima from Reflexivity’ option. Could you explain how the other options work for te Mask creation?

    -local maxima extraction with range lobes suppression
    -local maxima extraction with range and azimuth lobes suppression
    -regular grid, spaced dS by dL

    thanks

  • #1005

    periz
    Keymaster

    -local maxima extraction with range lobes suppression
    -local maxima extraction with range and azimuth lobes suppression

    these two options are useful for suppressing ghost targets generated by side lobes of the impulse response of the system (cardinal sine). in fact, side lobes could already be suppressed by filters used in the focusing process. so, it’s an advanced option useful only in particular cases.

    -regular grid, spaced dS by dL

    if you are working with full resolution data, don’t use this option. this option could be useful in case you work with filtered/multilooked data and you want to analyze a regularly spaced grid

You must be logged in to reply to this topic.