Authors: Justin Byrd*, Hexagon Geospatial
Topics: Remote Sensing, Spatial Analysis & Modeling
Keywords: machine learning, ERDAS IMAGINE, remote sensing, image classification, feature extraction, spatial analytics, spatial modeling
Session Type: Paper
Start / End Time: 10:00 AM / 11:40 AM
Room: Galvez, , Marriott, 5th Floor
Presentation File: No File Uploaded
With the current state of data deluge, machine learning algorithms are increasingly used for data analysis and information extraction. Machine learning, one of the hottest topics in the geospatial industry today, is the use of algorithms for creating knowledge from big data. These algorithms can be broken down into two main categories, Supervised and Unsupervised algorithms. Supervised algorithms learn from training samples and Unsupervised algorithms look at patterns and relationships without training data.
During this presentation, attendees will see demonstrations of machine learning with the latest release of the ERDAS IMAGINE Spatial Modeler, and walk-through the components to build spatial models from scratch. See feature extraction examples such as identification of rooftops, and land cover classifications using the Random Forest Classifier, which generates many decision trees. It is a direct graph where each node is a decision. Pixel values from different bands are used for making the decision. Once the decision tree is finalized, the entire image is processed. Users can view the results at each stage of the spatial model.