International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056 Volume: 03 Issue: 05 | May-2016 www.irjet.net p-ISSN: 2395-0072 © 2016, IRJET | Impact Factor value: 4.45 | ISO 9001:2008 Certified Journal | Page 1844 High Resolution Image Classification with Edge Detection based Segmentation and AdaBoost Manas Agrawal, Kapil Nagwanshi Manas Agrawal, manasagrawal.222@gmail.com, M.TECH Scholar, RCET Bhilai (C.G.) Kapil Nagwanshi, kapilk.nagwanshi@rungta.ac.in, Associate Professor, RCET Bhilai (C.G.) Dept. of Computer Science and Engineering, Rungta College of Engineering & Technology, Bhilai, Chhattisgarh, India ---------------------------------------------------------------------***--------------------------------------------------------------------- Abstract - Image classification is the most significant a part of digital image analysis. Classification of remotely detected information is employed to assign corresponding levels with relevancy teams with uniform characteristics, with the aim of discerning multiple objects within the image. A Geoeye-1 image is employed for experimental information. Firstly, the Geoeye-1 image is segmented, then abstraction, spectral, textural, color area and band ration options are designated. During this paper, image segmentation and AdaBoost technique is given and applied to object-oriented high resolution image classification. AdaBoost is an Adaptive Boosting machine learning meta-algorithm. AdaBoost classifier is intended considerately of confidence, support of well-mined rules. The visual comparison with the results of SVM and accuracy estimation validates the result of the projected approach. Key Words: high resolution images, classification, adaboost, image segmentation, thresholding. 1. INTRODUCTION Data mining is a vital research area in computer technology. It's far a computational process of figuring out patterns in massive information. Image mining is certainly one of crucial strategies in records mining, which worried in more than one disciplines. Picture type refers the tagging the photos into some of predefined sets. )t’s also consists of picture preprocessing, feature extraction, item detection, object classification, item segmentation, object class and lots of extra techniques. Image classification produces the accurate prediction outcomes of their goal magnificence for every case inside the data. It is a completely essential and challenging undertaking in diverse application domains, consisting of video surveillance, biometry, bio-scientific imaging, industrial visible inspection, vehicle navigation, remote sensing and robot navigation [1]. Resolution is that the capability of the detector to look at or live the tiniest objects clearly with distinct boundaries. Resolution depends upon the dimensions of the picture element. Usually, with any given lens setting, the smaller the dimensions of the picture element, the upper the resolution are going to be and also the clearer the item within the image are going to be. Picture having smaller picture element sizes may carries with it a lot of pixels. The amount of pixels correlates to the number of data at intervals the image. Image classification [2] is one altogether the foremost ways in which for information extraction from remote sensing image. Comparison with moderate-resolution, high- resolution image provides extra snug abstraction information, thus it's come-at-able to extract ground object extra accurately. The key issue of image classification is that the development of classifier. Classification of high resolution remote sensing information from urban areas is investigated. The foremost challenge in classification of high resolution remote sensing image information is to involve native spatial information inside the classification methodology. Image segmentation [4] is that the method of partitioning a digital image into multiple segments (sets of pixels, additionally called super-pixels). The goal of segmentation is to modify and/or modification the illustration of a picture into one thing that's additional meaning and easier to research. Image segmentation is usually wont to find objects and bounds in pictures. Segmentation of high resolution black and white geoeye-1 image is delineated. The regions were classified based on anyone of the 2 classes: Natural and Artificial. GeoEye-1 represents an additional step on the thanks to higher resolution capabilities for remote sensing satellites. This paper describes associate degree early experimental assessment of the accuracy of geo-referencing from GeoEye- 1 imaging. Thresholding is that the simplest methodology of image segmentation. From a gray-scale image, thresholding