Classification based land use/land cover change detection through Landsat images

Today, satellite data provide humans with immense information. This information, if used appropriately with technology will definitely yield us knowledge, which can be used for the betterment of mankind. This paper attempts to contribute in two ways a) classification of remotely sensed images to different classes and b) time sequence analysis of satellite images over a period of years. In the first study, for the purpose of classification, two non-parametric classifiers, Artificial Neural Network (ANN) and Support Vector Machine (SVM) are used. A comparison of both the classifiers is done using Kappa coefficient, and SVM is found to have outperformed ANN. The classification is done on the Landsat images for Kochi city, Kerala, India for the year 2014. In the second case, Landsat images of Kochi city from 2007 to 2014 are taken for study and a time sequence analysis is done. The images are classified into different classes and changes in the classes over the years are analyzed and it is realized that the highest loss of land use/land cover class has occurred to "Sparse Vegetation" and highest gain of the same has occurred to "Built-up" classes.