Diagonal Pixels?

[GeorgeClooney]

Hello Imagery People,

       I have recently been asked to work with NITF imagery that, somehow???,
is displaying diagonal pixels in arcmap. Do you guys have any experience with
this technique? I tried to run several conversions on them, however, all tools
seem to resample to square/rectangular pixels before processing. I can't find any
information on diagonal pixels. Is this a supported and accepted data strategy,
or is this most likely the result of a questionable implementation?

Thanks,

ModelGuy

[DarrenWiens2]

Is it possible that your data frame is displaying the data using an inappropriate spatial reference? As ArcMap projects your data on the fly, it appears to twist and distort your data according to the spatial reference. See below for a few examples using the same data and different spatial references.

[GeorgeClooney]

Hello Darren,

         Thanks you for your reply. Since the data is defined in WGS 84 and since adding it to a fresh open document still displays the pixels diagonally, I don't believe the diagonal pixels are the result of reprojecting on the fly, by arcmap. I think the only way this could be the case is if arcmap's WGS 84 definition was different than the WGS 84 definition within the NITF. I will compare these parameters as well. I was speaking to our remote sensing specialist and he said that it may be that whoever orthorectified the imagery was somehow able to avoid resampling the data. Still, I don't understand how arcmap would display the pixels diagonally unless it was accessing the transformation coefficients for display, somehow....

          Thanks for your help. Right now, I'm just trying to see if there are metadata tags in the NITF that will give me the transformation coefficients used for orthorectification. Please let me know if you think of any additional info. I think I need to look at the NITF format specification at this point. Do you know anything about NITF metadata tags, or if arcmap actually accesses these tags for display?

Thanks,

George

[RichardDaniels]

To start, NTIF Imagery is just a image format standard so does not tell us what sensor you are using. But in general, to get a true square or rectangular pixel throughout your entire image you would need to use a projection that supports meters or feet and the source imagery would need to have been reprojected into such a coordinate system.

If you have raw imagery in WGS84 the pixel size would be defined using longitude and latitude (probably at lower left corner of the image), in that case the "true length" along your east-west edge of each pixel would change as you moved north from the lower left corner of the image resulting in the stretching you see. Since the Earth is rotating while the image is taken you will also see a left-right shift in the location of the 1st pixel in each line as the sensor moves in relation to the Earth. That is why a area of interest is usually "clipped" from a larger image or mosaic to obtain a rectangular image that is used in a application.

Hope this helps.

Rich