Supervised Image Classification
Lab description - Students utilized unsupervised classification methods to derive Land Use Land Cover off of a satellite image. This week,the students were assigned to utilize supervised classifications to derive Land Use Land Cover (LULC) classification off of the spectral information contained in the Digital Numbers (DN) stored in remotely sensed imagery. Supervised classification differs from unsupervised methods in its uses of "training sites" (based on a priori knowledge often from ground observations of the information classes being mapped) to guide the classification of the image from spectral values into information (e.g. LULC) classes. It also differs from unsupervised classification in its use of statistics (as opposed to Euclidean spectral distance) to assign pixel values to information classes.
Student Learning Outcomes:
- Create spectral signatures and AOI feature
- Produce classified images from satellite data
- Recognize and eliminate spectral confusion between spectral signatures
STUDENT SPOTLIGHT AWARDS
The following student was chosen for their exceptional work on the Supervised Image Classification Lab assignment:
Gail Sease
What we like: Gail has been a hard working engaged student all semester and her work on the labs demonstrates her commitment to learning the concepts and techniques of remote sensing. We like that Gail explained the process of identifying training sites, how they relate to the spectral values, and the process for finding the signatures and spectral bands that have the least spectral confusion and therefore the highest potential of separability for each LULC class. Gail also identified one of the more effective band combos and her description of the spectral distance file assisted with the readers interpretation of her results. which she visually presented in her map. Way to go, Gail!
