Sahota Roseleen, Kundra Harish; International Journal of Advance Research, Ideas and Innovations in Technology © 2019, www.IJARIIT.com All Rights Reserved Page |202 ISSN: 2454-132X Impact factor: 4.295 (Volume 5, Issue 5) Available online at: www.ijariit.com A review on various approaches of remote sensing based satellite image classification Roseleen Sahota roseleensahota6@gmail.com Rayat Bahra Group of Institutions, Ropar, Punjab Dr. Harish Kundra hodcseit@rayatbahra.com Rayat Bahra Group of Institutions, Ropar, Punjab ABSTRACT Image Classification is the process that has been done for the extraction of valuable regions from the image. In this paper various approaches have been discussed that has been used for process of image classification. Supervised image classification and unsupervised image classification has been widely used for classification process. The classification has the major advantage that provides information about the various regions that are scattered over the satellite image. This has been used for prediction of percentage of various regions under different maps. Supervised classification approaches are based on machine learning algorithms that have been used for different types of prediction and decision making processes. On the basis of various approaches that have been discussed in this paper one can predict best approach for image classification process. Keywords— Image classification, SVM, Near Neighbor Based, PSO, BAT, MDC, Membrane Computing 1. INTRODUCTION 1.1 Image Classification in Remote Sensing Image classification is the process of extraction of valuable information from the remotely sensed images so that various regions from the images can extracted. In this processing of image classification various types of images have been used for generation of thematic maps over the different regions. An image has been divided into different pixel group for extraction of various land cover representation over remote sensing images through satellite. Land cover images divide images into different image representation regions that are urban area, forested area, rocky area, and water region and agriculture area. 1.2 Image Classification Techniques in Remote Sensing • Unsupervised image classification • Supervised image classification • Object-based image analysis Pixel is the smallest unit of the image that has been used for the representation of the image. Image classification process uses this pixel information and group of different pixels for generation of various groups from dataset. First two classification approaches are mainly used in the process of remote sensing image classification. Another approach that is object-based classification has been used for evaluation of the image regions based on different objects available in the images so that better classification can be done based on training input. 1.3 Unsupervised Classification Pixel available in the images is used for classification process so that these different pixels can be used for combination and represented as cluster. On the basis of cluster representation various groups of pixels have been divided into different classes. Various numbers of clusters have been used for formation of various brands of the image so that various regions can be extracted. Fig. 1: Unsupervised Classification Example The image classification software generates cluster. There are so many images clustering algorithm like K-Mean and ISO DATA. The use identified each cluster with land cover classes. The unsupervised classification image classification approach is used when no sample site exists. Unsupervised classification is that in which classification has been performed on the basis of different classification approaches that based on machine learning that does not utilize any training class for extraction of classes from the remote sensing images.