GIS4025 Lab 5: Supervised and Unsupervised Classification

-
This was an interesting lab. Most of it was completed in ERDAS Imagine. We took a high resolution aerial photo of the UWF campus, and used an Imagine function to break it into 50 image classifications. It was "unsupervised" because Imagine chose the classes and class breaks. We then manually went through the attributes table and used the inquire cursor to change every class to one of 5 classes: forest, grass, buildings/roads, shadows, and mixed. We then merged all the rows into these 5 categories and used the recode function to produce a new raster. Using the recoded raster, we then added an area column to the attributes, and calculated the total permeable and impermeable surfaces in the map.

In the next part, we did supervised classification, meaning that we selected map features manually by both drawing polygons, and using the grow from seed function and adjusting euclidean distance and neighbrorhood settings to use the program to infer the extent of features. From there, we used Imagine to classify the entire map according to the signatures we created.

We also looked at spectral data in histograms and line charts to determine where spectrally similar and different features differed the most in certain bands, to determine what bands would be the best for showing each feature distinctly.

Finally, we got a list of features from Germantown, MD, and did the process from scratch using what we learned:

Comments

Post a Comment

Popular posts from this blog

GIS 4006 Computer Cartography Intro Post

-
Hello everyone, After taking one class off, I'm on my second class of the GIS certificate program at the University of West Florida. I got interested in GIS after going to a regional GIS conference a couple of years ago, and I've been trying to learn everything I can along the way. I'd like to move into a GIS career after finishing this program. Our first assignment was a StoryMap, and I built one showing places I've lived. Thanks for looking! https://arcg.is/0mS1eD
Read more

GIS4035 Lab 2, Land Use/Land Cover

-
Image
This was not a particularly hard lab, but the main challenge was time management. We were given an aerial photo of Pascagoula, MS and had to classify the entirety of it to level II land use/cover categorizations. This took a lot of inferences and assumptions - we were discouraged from using other sources in the initial part, then checked our accuracy with Google Street View. I actually had a fairly high accuracy.
Read more

GIS 4035, Lab 4: Spatial Enhancement and Multispectral Analysis

-
Image
This was one of the more interesting labs that we have done. The first part was primarily concerned with fundamental skills in ERDAS Imagine, like using the chip image and subset tool, opening histograms, using the enquire cursor, and exporting data for use in ArcGIS Pro. We also accessed public domain US satellite data to analyze in Imagine. In the later part of the lab, we identified features according to criteria on the lab guide, such as features that caused a spike in pixels in a certain range of brightness in specific EMF bands, and had to use visual analysis and the enquire tools in Imagine to find them. Then, we created maps with combinations of RGB colors in different bands to show off which feature met the criteria. This was interesting, in part, because we had to think about color theory a little bit and why something bright in certain bands, when assigned colors in those bands, would produce the map effects that we got (i.e. if we assigned blue and green to bands where...
Read more