Lab 5 Unsupervised and Supervised Image Classification

 

In this lab we learned how to use Unsupervised and Supervised classification techniques in ERDAS Imagine. In this exercise we create signatures using Areas of Interest to show the software what each pixel was and to which class it belonged to so that we could create an image. Once that image was create we recoded the classes in a manageable amount of classes and corrected an spectral confusion using the mean statistical plots, distance maps, and histograms to determine areas that could cause an issue. We corrected this using different combinations of the spectral bands Red, Green, and Blue. And then created the image and maps above. 

Lab 04 Spatial Enhancement, Multispectral Data, and Band Indices

 For this lab we worked with both ERDAS Imagine and ARCGIS pro to gather satellite data. Then use it in identifying features using band indices and histograms to determine the reflection of light shown by the images. This also included using a variety of filters to help to identify features an highlight certain features so that they stand out more and can be accurately identified. Additionally we learned to gather data from online sources and work with bands to display and find certain features like in the images below.

Lab03

 For this Lab I worked with ERDAS Imagine and learned how to work in it and manipulate the data. For example I learned how to open it and edit it to fit it to the viewer.  I also learned about how the number of pixels can affect the different types of resolution of the raster and how to retrieve and manipulate the attribute table in ERDAS. Then I learned how to take an raster and bring it from ERDAS to ARCGIS and us the data gathered to make a map like in the image shown above. Additionally I learned how to calculate the area and display it in both software programs.

Lab 02 Ground Truthing

 For this lab we classified areas of the Pascagoula Mississippi based on level 2 Land Use and Land Cover classifications giving each classification a different color to distinguish it from other classifications. Then afterwards we check our work for using google maps and created 30 random points within the boundaries of the image using the create random points tool to check if my classification was correct via the street view feature.  Then we marked the points accordingly with correct points being green and incorrect being represented with red. We then measured the accuracy of my classification  and corrected the point in the attribute table if it wasn't. For my map I was able to get a 90% accuracy rating based off the points that were mapped.

 

Lab01 Visual Interpretation

 

In this lab we were tasked with identifying some of the basic features of images that are used to identify objects in these images. For example I used my own institution to determine to classify each of the different level of tone from very dark to very light and textures from very fine to very coarse. In the second map I used one of four strategies to identify objects such as Shadow, Association, Pattern, and Shape/size. For example I was able to identify parking lots, cars, water towers, and light poles using one of the strategies and additionally see how false and true color can differ on the same image.