2

I am looking to scan two separate rasters, capture the pixel information, then draw a polygon of the matching pixels.

I have two aerials, both are in different geographic areas (same projection and within about 1 km of each other). I am looking to determine if, between the images, there are matching pixel grids.

4
  • Do you have ArcMap, QGIS, or something else?
    – Taylor H.
    Dec 18, 2013 at 15:03
  • Arc suite and QGIS - arcsuite is at the editor (standard) level without any of the analyst extensions.
    – MDHald
    Dec 18, 2013 at 15:28
  • Since the two rasters don't overlap at all, I'm not sure how you would overlay/compare them. In essence, you want to place the two on top of each other and mark any cell between the two that has the same RGB value. So, these two rasters are identical in dimensions?
    – Baltok
    Dec 18, 2013 at 16:54
  • principally they are identical dimensions. i want to have one imagery feature and scan through a much larger image and search for an identical feature. you can imagine i have an image with a field, river, and sidewalk. represented as 1,0,2 the second larger image has many features and would be represented as 2,3,1 3,3,1 2,3,1 2,2,1
    – MDHald
    Dec 18, 2013 at 18:30

1 Answer 1

3

You haven't provided us with much information to try and answer your question...

Assuming your pixel values are numerical and your rasters each cover the same area / have the same cell size, an easy way to find out which pixels match is to divide one raster by the other (for example using ArcMap raster calculator). Any cells that match in the two rasters will have a value of 1 in the output raster.

6
  • hey, i have two aerials, both are in different geographic areas (same projection and within about 1 km of eachother). i am looking to determine, if between the images, there are matching pixel grids.
    – MDHald
    Dec 18, 2013 at 15:31
  • essentially, i want to scan an image (image 1) and look for similar features in a separate image (image 2), then match image 1 to image 2.
    – MDHald
    Dec 18, 2013 at 15:38
  • 1
    But you want to eventually have this information georeferenced, or does its location not matter? BTW scanning images and then matching exact pixel values is not recommended. You will need to generalize and looking for a range of values to find a "match".
    – Taylor H.
    Dec 18, 2013 at 18:32
  • 1
    Building on Taylor's comment: exact RGB combinations are rarely used to identify similar features because many things influence spectral reflectance (i.e. an identical object will have different RGB reflectances depending on sun angle, cloud cover, etc). Ranges of values on the other hand are the basis of image classification and can be identified automatically using an unsupervised classification method. Unfortunately, this is in the Spatial Analyst ext in Arc, but look at unsupervised classification options in QGIS (with a large # of classes) to get started.
    – JWallace
    Dec 18, 2013 at 22:09
  • @TaylorH. you are totally correct, in that i will be eventually georeferencing it.
    – MDHald
    Dec 19, 2013 at 20:08

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.