Module 9 - Unsupervised Image Classification
Unsupersived image classification is a type of automated digital image processing that focuses on spectral pattern recognition techniques to categorize remotely-sensed imagery. For our lab assignment, we performed two unsupervised classifications using both ArcMap and ERDAS Imagine.
For the ArcMap scenario, we used a Landsat satellite image. To create a classified image using this software, the following Spatial Analyst tools are required (in this sequence): the ISO Cluster tool, and the Maximum Likelihood Classification tool.
For the ERDAS Image scenario, we performed the classification using a high resolution aerial photograph of the University of West Florida campus, using the following tool: under the Raster tab > Classification group > Unsupervised button > Unsupervised Classification tool. We learned some of the important settings of this tool, such as the Maximum Iterations, Convergence Threshold, and the Skip Factor, which can all alter the calculation, accuracy, and processing time of the output. For reclassification, we learned to use the Swipe, Flicker, and Blend tools under the Home tab > View group; as well as highlighting techniques to assist us in distinguishing pixel classes. Then, we merged the classes using the Recode tool under Raster tab > Thematic button. And finally, we calculated the difference between impermeable/permeable surface types (by percentage) from the acreage totals of each class. My final map product is depicted at the top.
For the ArcMap scenario, we used a Landsat satellite image. To create a classified image using this software, the following Spatial Analyst tools are required (in this sequence): the ISO Cluster tool, and the Maximum Likelihood Classification tool.
For the ERDAS Image scenario, we performed the classification using a high resolution aerial photograph of the University of West Florida campus, using the following tool: under the Raster tab > Classification group > Unsupervised button > Unsupervised Classification tool. We learned some of the important settings of this tool, such as the Maximum Iterations, Convergence Threshold, and the Skip Factor, which can all alter the calculation, accuracy, and processing time of the output. For reclassification, we learned to use the Swipe, Flicker, and Blend tools under the Home tab > View group; as well as highlighting techniques to assist us in distinguishing pixel classes. Then, we merged the classes using the Recode tool under Raster tab > Thematic button. And finally, we calculated the difference between impermeable/permeable surface types (by percentage) from the acreage totals of each class. My final map product is depicted at the top.
Comments
Post a Comment