Color-based SAR image segmentation using HSV+FKM clustering for estimating the deforestation rate of LBA-ECO LC-14 modeled deforestation scenarios, Amazon basin: 2002–2050

Abstract

Image segmentation is an essential process for image evaluation tactics. Synthetic aperture radar (SAR) image segmentation can be implemented for a vast magnificence of various problems. Due to the existence of speckle noise in SAR images, the edge attributes of the images are unknown. This leads over or under segmentation and produces poor quality of segmented results. In this paper, the deforestation rate of Amazon, South America, has been determined by using the color space-based SAR image segmentation. Since the fuzzy K-means (FKM) clustering technique is robust to speckle noise, it is combined with different color spaces for better segmentation. The proposed method has been compared with different existing methods, and it gives better segmentation results with a segmentation accuracy of 99.5%, Jaccard index of 99%, recall of 99.5%, and specificity of 99.7% with the minimum error rate of 0.004. The segmentation accuracy of 14.9% and the Jaccard index of 10.3% have been improved when compared with FKM clustering technique. This paper suggests that HSV+FKM is a suitable technique for the segmentation of LBA-ECO LC-14 modeled deforestation scenarios. Brazil’s National Institute of Spatial Research (INPE) discovered that the span of Amazon could be diminished to 50% by 2050. This paper also predicts that the Amazon rainforest will be reduced to 43.26% at the end of 2050.