Pixel-based Classification of Multispectral Remotely Sensed Data Using Support Vector Machine Classifier

Abstract: Image Classification is an important task within the field of computer vision. Image classification can be defined as processing techniques that apply quantitative methods to the values in a digital yield or remotely sensed scene to group pixels with identical digital number values into feature classes or categories. The categorized data thus obtained may then be employed to create thematic maps of the land cover present in an image. Classification includes Determining an appropriate classification system, selecting training samples, image pre-processing, extracting features, selecting fitting classification approaches, post-classification processing and accuracy assessment. The objective of this study was to evaluate Support Vector Machine for effectiveness and prospects for pixel-based image classification as a modern computational intelligence method. SVM is a classification technique based on kernel methods that has been proved very effective in solving complex classification problems in many different application domains. In the last few years, SVM gained a significant credit also in remote sensing applications. SVMs revealed to be very effective classifiers and currently they are among the most adequate techniques for the analysis of last generation of RS data. Keywords: Image classification, Support Vector Machines, Pixel-based, Multispectral, Remote sensing.