Knowledge Satellite Images and Laser Data

One of the applied studies in the field of object-based analyses is the extraction of urban features. Buildings are one of the most important urban features in large scale maps, so the process of identifying and extracting these features is especially important in the location-based processes of the passive defense. Due to the variety of spectral and geometric properties of these types of features, their identification in different study areas is associated with some problems. In this study, using an object-based analysis and extracted features from laser data and aerial photos to identify buildings with a sloping roof are discussed. In the first step, after the segmentation process, the separation of high and non-complicated features is done using the slope and gradient direction. In the next step, by extracting the geometric and conceptual features, the separation of trees and buildings from each other is done. In the final step, the reconstruction of the lost buildings is performed by morphological operators. The combination of two categories of input data at the decision level leads to benefits from the characteristics of both data, and each of them can cover other problems and shortcomings. In the proposed method, a rule-based strategy is based on the production of geometric and conceptual features and the use of a multi-stage approach. Finally, the overall accuracy of the building category identification is 87%, and the kappa coefficient is 81%. The results demonstrate the high capability of object-based methods to identify urban features, such as buildings, by combining Lidar data and aerial photos despite the diversity of urban environments.