Geographical analysis

Abstract

Abstract Terrain and land-use analysis are an integrated component of various planning and policy decisions. Classification based on statistical methods like supervised and unsupervised methods provide results with limited accuracy because of their inherent limitations of dependence on radiometric values. Rule based methods for object based classification for feature extraction produces better results for high resolution imagery. This paper intends to compare advantages and accuracy of using object based image classification techniques with such legacy classification methods. It also explores possibility of applying object based techniques for terrain / land-use feature extraction at two similar terrain sites using same rule set. The resultant outputs are characterized and compared from the perspective of defence operations in unfamiliar terrain at different locations.

Keywords: Classification, Object based classification, Supervised, Unsupervised, Terrain Analysis.