Classification Analysis of Tippecanoe County Amphitheatre

 Classification Analysis of Tippecanoe County Amphitheatre

Kendrick Wittmer

Date: 5/7/2025

    The objective of this assignment was to perform both unsupervised and supervised classification of the Tippecanoe County Amphitheatre, and then use the classification process to perform further analysis by identifying permeable and impermeable surfaces and crack detection. 

    After resampling to a smaller cell size, the classification wizard was used to carry out the objectives of this assignment. For more information on specific settings and the setup of unsupervised classification, refer to the previous post regarding object-based classification. 

    The result of the unsupervised classification was far from desirable, however the learning goal of this assignment was not necessarily to produce high quality classifications but to gain a base familiarity of the process. Looking back, the main issue stemmed from having way too many classes. The maximum number of classes for this application should probably be set to 4 or 5.

    While the unsupervised classification turned out poorly, after grouping the 8 different classes into only permeable and impermeable, we can actually begin to see some pattern develop. Some of the roads and the building are visible. Again, adjusting the number of classes in the unsupervised classification may help produce a better product. 

    The next step was to perform supervised classification of a small portion of the area with the purpose of crack detection. Only three classes were created for this portion: Pavement, Vegetation, and Cracks. For supervised classification we must select our own training samples to input in the classification wizard. I selected fairly broad samples for the grass, slightly small samples for pavement, and for the cracks I zoomed in and selected only a few training samples each consisting of around 8-10 pixels. This yielded a very accurate classification of the cracks in the pavement. 

    This assignment made clear the uses of classification for analysis in the UAS industry. At first I thought classification was just a different form of digitizing which is performed by the computer, however now I see how it can be used to gain a broader understanding of fine details within a dataset.