Comparison of advanced classification techniques in remote sensing

Studies on remote sensing have become inescapable for almost all disciplines along with the advances in aviation and space technologies. Such as defence, city planning, forest services, naval surveys, mining etc many disciplines use remote sensing technique. One of the parameters that makes this discipline important is classification studies. This parameter is divided into two: supervised and unsupervised. Multi and hyper spectral images are processed mathematically in this parameter and they help us become informed regarding land cover. Basic definitions used in remote sensing are discussed in the first section of this study presented. The mathematical analysis of the support vector machines, artificial neural networks and decision trees, which are also known as further classification algorithms, are included in the second section. Urla semi-island, Izmir, Turkey has been classified by using LANDSAT 8 images by the help of these algorithms in the third and the final section of this study. Classification findings have been taken into verification testing by means of confusion matrix which is formed under favor of the spectral reflection values of the training classes. Besides, which algorithm has an impact on what kind of classes has been determined by the help of confusion matrix.