sarproz-group

ALOS Slaves Extraction

This topic contains 2 replies, has 2 voices, and was last updated by  periz 2 years, 9 months ago.

  • Author
    Posts
  • #764

    matusbakon
    Participant

    Hi,

    I have problem with Slaves Extraction of ALOS images: “Size inputs must be integers”. Does it mean that image is corrupted? I’ve tried to corregistrate the set of 15 images before (in old version), get succesfully through Slaves Extraction but the corregistration went wrong – corregistrated slaves were shifted directly in the image where Slaves Extraction is stopping now. Does any of you have hints on corregistrating ALOS images? There seem to be huge shifts (solved by increasing Delta parameter and increasing reference window and search pixel) and there are no orbits (ticking from Coarse Xcorr), should I consider anything else?

    Thanks in advance for your response.

    Matus

  • #765

    periz
    Keymaster

    1. “Size inputs must be integers” never happened to me before. please let’s have a remote session, if the image is corrupted we need a better check with consequent message.
    2. “corregistrated slaves were shifted directly in the image where Slaves Extraction is stopping now” don’t know what you mean
    3. Higher Delta and Coarse Xcorr usually solve the problem of big shifts. However, should your image be highly stretched, you may increase also the search pixels
    4. to attach log and mat files I suggest to tar them first

  • #766

    periz
    Keymaster

    Good we have analyzed and solved all matters.

    1. if in a dataset of 15 images there are 2 or 3 bad images with high shifts, it is useless to process all of them with non-standard parameters: firstly, one has to coregister the dataset with standard parameters, after that one can deal with the faulty images one by one searching for the proper way to coregister them. The sw allows adding images to an existing dataset.

    2. when dealing with mixing of multiple polarizations, the software automatically chooses the Master image among the single pol ones (if available). In this way, the maximum resolution is kept and images with multiple polarizations are oversampled. You are free to change it and choose a different master, but then, if you chose a multi-pol one, you should manually choose an oversampling factor equal to 2. I’ll add a check and a reminder on this

You must be logged in to reply to this topic.