International Journal of Recent Technology and Engineering (IJRTE) ISSN: 2277-3878, Volume-8 Issue-3, September 2019 5746 Published By: Blue Eyes Intelligence Engineering & Sciences Publication Retrieval Number: B2607078219/2019©BEIESP DOI:10.35940/ijrte.B2607.098319 Abstract: Amazon region plays a very substantial role within the worldwide energy, hydrological cycle and carbon balance. The total area of Amazon forest is 5.5 million km 2 . About 7.3% of the Amazon was deforested between 1976 – 2003 and further 2.6% between 2000 and 2010. PRODES (Amazon Deforestation Monitoring Project) oversees deforestation over satellite images of the categorical deforestation in Amazon. The annual deforestation rate is predicted depends on the desertification expansion identified in every satellite image integument the Brazilian region known as legal Amazon. According to Brazil's National Institute for Space Research (INPE) and Food and Agriculture Organization (FAO), till now totally 768,935 km 2 were deforested in Amazon Basin. This paper proposes a segmentation approach that enables the estimation of the deforestation rate on tropical forest clearing in Amazonia, South America in the recent, past (2002–2010) and in an exceedingly future (2020–2050) land cover change in the LBA-ECO LC-14 Modelled Deforestation Amazon Basin scenarios. Index Terms: Amazonia, Deforestation, Segmentation, Land Cover, Fuzzy k-means. I. INTRODUCTION Synthetic aperture radar (SAR) images are used increasingly more in various fields [3,10,13] which includes geosciences and climate change research and detection, atmosphere and earth system monitoring, 2-D, 3-D and 4-D mapping, security-related applications and planetary exploration [1]. It is specific in its imaging capabilities such as side-looking imaging geometry, high-resolution, day-and-night time and climate autonomous. SAR image is captured in such away, that the radar framework transfer radar pulses with excessive power on the objective zone of earth and records the reverberation of the back dispersed signal in a sub sequential manner. Deciphering, aperture radar images are substantial part in evaluating the Earth’s surface attributes of the imaged area [14]. SAR image segmentation is imperative in radar image processing. The primary intention of segmentation is segregation of an image into locality sub areas of dissimilar Revised Manuscript Received on September 23, 2019 Kalaiyarasi Murugesan, Electronics and Communication Engineering, Kalasalingam Academy of Research and Education, Krishnankoil, India. Perumal Balasubramani, Electronics and Communication Engineering, Kalasalingam Academy of Research and Education, Krishnankoil, India. Pallikonda Rajasekaran Murugan, Electronics and Communication Engineering, Kalasalingam Academy of Research and Education, Krishnankoil, India. attributes [12]. Mostly aperture radar images are affected by crucial speckle noise. Due to speckle, segmentation of radar image is further difficult than that of other images. Segmentation can be approximately detached into region-based, edge-based methods [7,15], morphological methods [9], clustering methods [2,8] and markov random field [4]. In order to segment the SAR images effectively, nonlocal fuzzy clustering algorithm with edge preservation was proposed. This technique was effective for the SAR images with fewer speckle. In the case of more speckle noise, some characteristics of original image could be lost [5]. A kernel FCM algorithm with pixel intensity and location information was proposed for segmentation, it depends on the geographical and anxiety distances of all adjoining pixels concurrently. Spectral Clustering (SC) is also suitable for image segmentation [6,11]. SAR image generally has multiple regions of interest (ROI) and is encapsulated with non-additive multiplicative noise has mysterious distribution. Number of effective clustering algorithms have been used for SAR image segmentation, among which the k-means clustering is more robust to speckle noise [12]. In this paper, Fuzzy K- Means clustering technique is used for segmenting forestry land cover. The rest of this paper is standardized as follows. K-means clustering based segmentation is briefly explained in the next passage. Furthermore, the segmented results by employing the fuzzy K-means clustering on LBA-ECO modelled deforested amazon images are presented. Decisively, a few denouements and pertinent deliberations are given. II. FUZZY K-MEANS CLUSTERING The K-means clustering is a simple straight forward clustering methodology which uses reiterative method of grouping things into k clusters, here k is the number of pre-picked groups. The grouping is carried out through limiting the Euclidean distances between items and also the relating centroid. A centroid is the centre of mass of a geometrical object of uniform density. K-means approach is an unsupervised clustering calculation that picks the cluster center insightfully and it contrasts with the image pixels depends on their intensity value, characteristics and calculate the Euclidean distance. The image pixel values which are comparative toward the cluster center are allotted to the cluster having the same cluster center. New k' cluster center are calculated and therefore Estimation of Deforestation Rate for LBA-ECO LC-14 Modeled Deforestation Scenarios, Amazon Basin: 2002-2050 using Fuzzy k-means clustering Kalaiyarasi Murugesan, Perumal Balasubramani, Pallikonda Rajasekaran Murugan