I use NAIP imagery (primarily in QGIS/GRASS) for agricultural analysis. I want to separate out objects such as grain silos, farms, buildings, &c. using segmentation and classification.

Is this possible with NAIP's somewhat lackluster resolution?

I know with LANDSAT imagery I could pansharpen, but I don't have that luxury with quad-band.

  • You can certainly segment NAIP imagery although to achieve the detail in the classification you want may not be possible. You are also confusing land cover with land use. Its unlikely you'll be able to classify farms and buildings separately, given that they are basically made of the same land cover types – Nathan Thomas Jun 12 '18 at 18:09
  • I think I read you on the land cover type issue. I'm going to keep plugging away with segmenting and devising training samples and see what I get, if I have to find higher-res imagery, so be it. Thank you Nathan! – instructorjack Jun 12 '18 at 21:24
  • A quick lit review would serve you well. Look at what schemes have been used previously (with NAIP imagery) and how successful they were at identifying your classes of interest. For example, St. Peter et al. (open access) used NAIP imagery and a softmax NN for probabilistic classification of shadows, pavements, buildings, tree crowns (light, dark, & grey), shrubs, grass (green & dry), water, burned areas, bare ground, & crops (no affiliation - just the first recent open access NAIP-only classification I found). NB: pixel-based rather than GEOBIA. – Dave G Jun 12 '18 at 21:56
  • Are you referring to lackluster spectral or spatial resolution? – Aaron Jul 29 '18 at 2:50

Your Answer

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

Browse other questions tagged or ask your own question.