A Novel Approach for improving Post Classification Accuracy of Satellite Images by Using Majority Analysis

Abstract

In past one year, due to climatic changes and some anthropogenic activities, the forests of Uttarakhand are burning. To identify the damage caused by the forest fires, an area of Nainital district has been taken for the study. Multi temporal Landsat 7 images were taken from April - 2020 and April – 2021. This paper shows a novel approach to increase the accuracy of the classified image. The Support Vector Machine classification is first done and then to improve the accuracy of the classified image, a post-classification technique called Majority Analysis is applied. This method helps to classify the unclassified pixel and it also smoothens out the boundary of the classified pixels, leading to higher accuracy rate. The classification accuracy has improved significantly for April 2020 and April 2021 images from 89.35% to 98.71% and from 88.52% to 99.76% respectively. The change detection study showed a drastic increase in the barren land due to the forest fires and on the contrary, the forest, scarce forest and the shrub land areas have decreased.