Image Classification for Remote Sensing Using Data-Mining Techniques

Abstract

Remote Sensing engages electromagnetic sensors to measure and monitor changes in the earth's surface and atmosphere. Remote Sensing Satellites are currently the fastest growing source of geographical area. Using data-mining techniques enables more opportunistic use of data banks of remote sensing satellite images. This thesis focuses on supervised and unsupervised classification, the two data mining techniques on the high resolution satellite Imagery from satellite IKONOS and satellite LANDSAT taken of the area around Kent State University, Ohio. The image was classified into ten distinct class: 1) Water, 2) Forested, 3) Agriculture, 4) Urban Development, 5) Vegetation1, 6) Vegetation2, 7) Vegetation3, 8) Vegetation4, 9) Grass, 10)Road. ERDAS Imagine was used in manipulating the images and creating the classification and analysis. The result obtained in form of accuracy helps to decide which image and classification technique is better to identify geographical patterns related to land use.