Comparison between standard classification methods for best suitability in urban scenario

Satellite image is a picture of the terrain taken from far above the earth. It therefore has as many features as the terrain beneath. It becomes a challenge to categories these features correctly on the image. Categorizing the features into various classes is termed as classification of the satellite image. When the classification is done digitally, different geostatistical algorithms or classifiers, as they are called, may be used for the same terrain. A classifier may suit a particular geographical location and at the same time may not be appropriate for a different geographical location. It becomes a matter of study to find out the best suitable algorithm for classification digitally for a particular terrain. In this research, three classification techniques have been used, namely Deep Learning, OBIA classification and supervised classification. These three classification techniques were applied in this study to generate a proper and accurate land use map extracting urban area data based on the maximum accuracy. Since different classification techniques are applied on different areas in order to gain maximum accuracy hence, in order to extract urban area data in this study, three classification methods have been compared to check which one is most accurate and suitable.