Title: Accuracy analysis of various classification algorithms for used land

Authors: N. Suresh Kumar; M. Arun

Addresses: School of Computing Science and Engineering, VIT University, Vellore, Tamil Nadu, India ' School of Computing Science and Engineering, VIT University, Vellore, Tamil Nadu, India

Abstract: This research work describes the region-based approach for satellite image classification to extract land use details in the Vellore District of Tamil Nadu, India. The objective is to find the greenery and used lands of the study area by classifying the satellite imagery using fuzzy-based, K-nearest neighbourhood, support vector machine classification methods and spectral information of a LANDSAT satellite image. The LANDSAT image is applied along with the image processing algorithms to get classified image. These algorithms are implemented and the objects are identified. These identified objects and ground truth values of study area are compared. By comparing producer's accuracy, user's accuracy, omission error and commission error, the overall accuracy is calculated and the algorithm which gives the better performance is identified for LANDSAT image of the study area, Vellore District, Tamil Nadu, India.

Keywords: fuzzy logic; K-nearest neighbour; fuzzy k-NN; support vector machines; SVM; classification accuracy; India; satellite image classification; land use; LANDSAT images; image processing.

DOI: 10.1504/IJENM.2016.077527

International Journal of Enterprise Network Management, 2016 Vol.7 No.2, pp.113 - 132

Received: 13 Nov 2014
Accepted: 06 Jan 2015

Published online: 05 Jul 2016 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article