Classification of urban environments using feature extraction and random forest

Multisource remote sensing data provide information of high relevance for classification and climate studies in urban areas and are of particular interest for regional and global climate science. To classify the urban environment using predefined Local Climate Zones we propose a method using feature extraction, image segmentation and decision trees. The method extracts features and segments from the multisource data. Later objects are selected to create decision trees using the random forest algorithm. Finally the images are classified by the earlier generated trees. The multispectral images were from Landsat 8 and Sentinel 2 resampled to 100m. Visual analysis and quantitative testing of results show the effectiveness of our method.