Principal component analysis based classification of settlements in satellite images

Abstract

The objective of this research is to use satellite images for the classification and identification of settlements. Satellite images are used in this research. A wide area is covered in a single satellite image and it contains enormous information therefore satellite images can be used for many useful purposes, like classification of objects and land cover classes, change detection and population estimation etc. Many small settlements are usually scattered in remote areas. These settlements are very important in many aspects for a country such as; the data can be used developmental planning and the accurate censuses statistics of a country; moreover this data can be used for damage detection in case of any disaster. Furthermore it can help estimate etc. This is very laborious and expensive task to visit each and every place but this work can be carried out very easily and affordably using satellite imagery. To obtain the desired information from satellite images many different classification algorithms are used. The algorithms are broadly categorized into supervised and unsupervised classification. The selection and result of these algorithms depend upon the nature of the problem to be analyzed. In this work Principal Component Analysis (PCA) is used to extract the feature or signature in the shape of Eigen vectors. These Eigen vectors are used for classification on the basis Euclidean distance. Comparing the actual results with the reference map checks the accuracy of the system.