Presenting and comparing the object based image analysis and standard image analysis for change detection of forest areas, using low-resolution satellite imagery

Abstract

The objectives of this study were to present and compare methods for change detection of forest areas in low-resolution satellite imagery, using object based image analysis and standard image analysis techniques (pixel based). Considering that huge area in Serbia is covered with forest, this study is of great importance for detecting and tracking changes over time. The data used in this study were optical remote sensing satellite images Landsat TM (22-08-1986) and Landsat ETM+ (23-06-2002). These images were processed in order to classify and identify four main land cover classes, namely: forest, agricultural, urban and water areas. The images are analyzed and the process of change detection was applied on forest areas. The results show the type and the place of landuse changes occurred in these main four classes during those years. Change analyses were performed based on object image analysis and standard image analysis techniques on the other hand. The research reported here shows that the object-oriented classification method is most accurate when it comes to the classification of land cover. The research also shows that the method of post-classification on object classification results is the most accurate method of change detection in forest areas. © SGEM2011 All Rights Reserved by the International Multidisciplinary Scientific GeoConference SGEM.