Authors: Charlie Frye*, ESRI, Inc.
Topics: Population Geography, Applied Geography, Geographic Information Science and Systems
Keywords: Population, Landsat, Landsat8, Texture, Image Processing
Session Type: Paper
Start / End Time: 2:00 PM / 3:40 PM
Room: Studio 10, Marriott, 2nd Floor
Presentation File: No File Uploaded
The fifteen-meter resolution of the Landsat8 panchromatic band offers sufficient texture in areas of human settlement to statistically derive a weighted dasymetric footprint of where people live. However, there are complications to working with Landsat8 data that must be overcome when producing a global footprint of human settlement. In this presentation, the focus will be on using image texture, i.e., local statistical variation or patterns, as the basis for classifying a raster cell as human settlement. The methods for detecting texture, selecting Landsat8 scenes, and eliminating statistical variations between scenes will be discussed. We will also discuss the advantages and challenges of applying this method to finer resolutions of grayscale imagery, and to color imagery.