Extracting roads from remote sensing data sources is an essential task for urban surface feature analysis. Automatic feature extraction from remote sensing imagery reduces the need for people to personally perform time-consuming field survey tasks. It is not only time-saving, but also easy to update in a timely way. In recent years, various sources of remote sensing datasets have been made available and are widely used in feature extraction. This allows researchers to construct methods to extract certain features from the imagery effectively. This research explores the use of IKONOS data combined with LiDAR data for urban road extraction in Indianapolis, Indiana. The data resource, extraction process, effectiveness, and applicability of three extracting methods are compared. Although combined dataset has higher requirement for a data resource, its cooperation with object-oriented method obtains the best result among the three. Four accuracy measurements are used to evaluate the extraction results. By comparing these four measurements it shows that the object-oriented method with combined dataset greatly improves the extraction accuracy and quality. |