I am working with Landsat 8 images and some of the bands have striping, e.g.: enter image description here

I've been going down a rabbit hole by looking for a solution for this. Does anyone know a python library that could be used for this?

  • 1
    I can't offer much hope but landsat.usgs.gov/… says you're not the only one with this problem. According to the linked page the problem is at the sensor so would need some quite spectacular coding to even detect the banding programatically. How does the banding differ from other image bands? That is can another band (or combination) be used to normalize your data? What does it look like on different days? – Michael Stimson Jan 15 at 5:10
  • 1
    You could try low pass filtering which would reduce quality or maybe try a fourier transformation and mask out the problematic areas in the magnitude. This however assumes a consistent striping pattern. – Carlos Jan 15 at 10:17
  • Which bands have you been observing this phenomenon? – Aaron Jan 25 at 15:55
  • @Aaron, Mostly on the 6th (SWIR-1), which is the most important one for my analysis. Also get a lot on the TIR ones (10&11), but I don't need those. – A_R Jan 26 at 2:21

The striping you are seeing in the imagery is common in satellite imagery and unfortunately, there is no reliable way to automatically identify and remove them, without causing significant loss of information in other areas / cases.
As such, the general approach is to minimize impact from the stripes by relying on less sensitive applications, since the magnitude of the noise is usually only a few DN compared to the other areas.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.