A Comparison of Classification Techniques for Glacier Change Detection using Multispectral Images

Publication date: Available online 4 May 2016 Source:Perspectives in Science Author(s): Rahul Nijhawan, Pradeep Garg, Praveen Thakur Main aim of this paper is to compare the classification accuracies of glacier change detection by following classifiers: sub-pixel classification algorithm, indices based supervised classification and object based algorithm using Landsat imageries. It was observed that shadow effect was not removed in sub-pixel based classification which was removed by the indices method. Further the accuracy was improved by object based classification. Objective of the paper is to analyse different classification algorithms and interpret which one gives the best results in mountainous regions. The study showed that object based method was best in mountainous regions as optimum results were obtained in the shadowed covered regions.

Source: Perspectives in Science - May 4, 2016 Category: Science Source Type: research