A Comparative Study of Various Methods for Forest Change Detection

Abstract

This proposed research gives an insight over the forest change detection using the satellite images for a decade starting from 2007 to 2017. For this purpose, a comparative analysis of different vegetation change detection methods utilized. This work also utilizes the Bhattacharya distance for the proposing of an enhanced forest change detection model which is based on pixel based method. Initially, two satellite images taken into consideration from the same geographical area. Then, the pre-processing is applied on these images by the filtering approach. After the pre-processing stage, the feature extraction carried out using different vegetative indices Viz. RVI, NDVI, EVI, WDRVI, GARI, Green Chlorophyll Index (CI Green), Green Leaf Area Index (GLAI). Here, the forest area segmentation was performed by using LBPS, USNN & DNN. The segmented results from both the images give the change detection in forest cover with Bhattacharyya distance. For the year 2007 the variance range was in between 0.245987 to 3.593802 and for the year 2017 this range varies from 0.942016 to 10.41409 for different vegetative indices.