Remote Sensing Student Spotlight

GIS4035 Photo Interpretation and Remote Sensing, Mr. Brian Fulfrost

Unsupervised Image Classification

In this week’s lab, students learned to perform unsupervised classification using both ArcGIS and ERDAS Imagine. Along with running the classification process, students were asked to determine what feature the new image classes represent (reclassification) and then simplify many classes into a few types of features (recoding). The map submitted was a reclassified and recoded image of the UWF property including text identifying impermeable and permeable surfaces as percentages.  

Student Learning Outcomes:

• Perform an unsupervised classification in both ArcMap and ERDAS 
• Accurately classify images of different spatial and spectral resolutions 
• Manually reclassify and recode images to simplify the data 

STUDENT SPOTLIGHT AWARDS

The following student was chosen for their exception work on the Unsupervised Image Classification assignment:

Justin Coryell

About Justin: Justin is originally from Arizona, but currently lives in Mississippi with his wonderful family. He is an Aerographer's Mate Chief in the Navy and has been in for 15 years. An Aerographer's Mate is a name applied for Weatherman, Chief is the rank. Besides Mississippi, Justin's duty stations have been all over the world; Yokosuka, Japan, Naples, Italy, and San Diego, California, along with deployments to South Carolina, Western Pacific, Indian Ocean and the Arabian Gulf. He received his bachelor's degree from Mississippi State University in Broadcast/Operational Meteorology in 2007 and wanted to do more with GIS. So, Justin joined the UWF GIS program in January 2013, and has since loved every minute of it. Welcome to the spotlight, Justin!

What we like: In this week’s lab, Justin’s reclassification and recoding was exceptional.The combination of both Justin's (a) description of the process and technique of unsupervised classification, and his (b) map, which was well designed and clear,  demonstrated his high level of comprehension of the week's material on automated image classification methods. Justin also provided a brief critique of unsupervised classifications  including (1) the limitations of unsupervised classification of spectral values to distinguish every pixel as pervious or impervious cover classes, and (2) issues with the pixel values of the imagery (e,.g. shadows). Although the vast majority of students did very well on this week's lab, Justin's map and description provide the best overall blog posting.