Image Classification in ArcGIS Pro - The Basics

[Music] hey everyone Chris here and in today's video I'll be talking about image classification within ArcGIS throw specifically unsupervised image classification it's gonna be very basic but I hope you stick with me because it's pretty cool at the end here we go alright here I am back at the office and I'm gonna show you how to use classification unsupervised classification using ArcGIS Pro on color infrared or false color imagery now I'm gonna show you the very quick method you can tweak all the parameters if you like but I'm just gonna show you so you can get started on classifying your color infrared imagery and keep this in mind next video I'm gonna show you something very bad about classification anyways here we go here we are in ArcGIS Pro I have loaded up my color infrared imagery that is three bands in that imagery that is near-infrared red and green the resolution in this case it's considerably lower than my high resolution I'm gonna zoom in to show you the resolution on the color infrared in this case is 1.5 meters by 1.5 meters Wow my high resolution is 0.3 by 0.3 meters so there is a massive difference in the resolution in the high resolution we get in see shadows and the color infrared we cannot I'm gonna zoom out I'm gonna zoom out to show my entire area to start our unsupervised classification of the color infrared or false color imagery depending on your nomenclature you first have to click on the imagery itself then come up to the ribbon click on the imagery tab and then click on the classification wizard icon a new darker appears image classification wizard now there are a lot of settings in through here but what I'm gonna do is I'm only gonna I'm gonna keep a lot of them defaulted else I'll say which ones I prefer but we're going to go for default many defaulted for now so first is the classification method you can choose between supervised and unsupervised I'm going to go for an unsupervised there's a classification type pixel-based an object-based I prefer object basis because when I've done them in the past it seems to come out better but your choice when you come to your own unsupervised classification now the classification schema again I'm leaving this as default this is going to be the it's the N LCD that is the National Land cover database schema I don't have any of these other options so I'm just going to click on next so what is happening now is it's creating a segmented preview so here's some more values that I'm just going to default for now this is this is the segmentation settings again defaulted next so now it's happening it is running a set creating a segmented preview of the settings you just chose segmentation preview is completed it looks a bit different than your color infrared of course this preview is not saved for instance if you closed your ArcGIS pro right now you would lose this segmentation not the in the world depending on the size of your room imagery you can get back here fairly quickly back over to our darker this is now a training docker I'm gonna leave all of these values as defaulted I will say normally I do instead of taking five maximum number of classes I usually do 11 that's because I want to see if there's more classification possibilities than just the five presented in the default setting I will leave it default for now next we click down on run and after a few minutes and the magic of Adobe Premiere Pro we have a segmented and a previewed classification for our color infrared imagery next is to classify within the next is to classify within the image classification wizard again we're gonna leave this optional for now and click run now you've got an unsupervised classification which we can adjust now based on based on the nlcd or national land cover database as you can see it's not quite perfect as you can see we this is where we can make some decisions now I know this is Road in through here so that's this color hopefully you're not colorblind and we're going to call this developed so you click on a sign and developed and again I know this is a road so we click on it is accept that that whole classified area as developed I also know this area is for basis I now select this area as herbaceous you will want to check your classifications against probably in this case since we really only have one other set of imagery and data against your or against my high-resolution imagery zoom and have a look I know these are spiky trees and what are spiky trees they are coniferous so we're gonna click this as evergreen I'm gonna zoom in and look these are fluffy trees so those are deciduous now we've got one last class in this case to classify and of course it's a color I can't see so let's see if we could change it we cannot one last classification to name fortunately I picked some not-so-great color or it picks them not-so-great color so it's green luckily I can double click on the old classification and now I can zoom in and to see I have a look at what we're possibly looking at and if I click back and forth I'm gonna have to say it's deer I have to just say it's mixed forest yeah we're gonna call it mixed forest for now so come back we're gonna call this mixed forest and there we have our unsupervised classification named by a person and we're going to go and now we've got a unsupervised classification named by a person so I'm gonna click on next now we can do another reclassification I am not gonna bother I'm just gonna say hey this looks fantastic now we could do edits on the classification just in case there was mistakes made or that the imagery wasn't perfect and we knew where things had to be changed in this case I'm just gonna leave this default again and I'm gonna change the name that makes sense in this case it is WV 275 3 it's coming in for red and I'm gonna call this on soup classification version 0 1 because I may want to come back and try a whole bunch of different parameters instead of the defaults which I just used and then I click run there we have our final unsupervised classification of our color of red or false color imagery it's not a perfect setup and it was mostly defaults but you get the gist of where you can go from here now there are a lot of problems with the way I just did that unsupervised classification in fact it's quite bad because I wasn't using any ground true thing and I was really only using one set of imagery that being the high-resolution imagery as a cross reference for my classification in future videos I'll talk about what you can do to make sure that your classification is better thanks for watching if you like this video please give me big thumbs up subscribe my channel add me to your LinkedIn or even better share my videos to your networks until next time I'm dr. Chris [Music] [Music]

Channel: Dr. Chris Geoscience

Views: 4,448

Rating: 4.9658117 out of 5

Keywords: Geophyics, Geology, Geoscience, GIS, Geographic Information Systems, Science, Environment, Seismic, Magnetics

Id: zkNWjme-BOs

Length: 9min 49sec (589 seconds)

Published: Fri Oct 18 2019