ABSTRACT
CONTENTS 9.1 Introduction ............................................................................................... 163 9.2 Data ............................................................................................................. 165 9.3 Methodology.............................................................................................. 166
9.3.1 Preprocessing................................................................................. 167 9.3.2 Data Reduction.............................................................................. 167 9.3.3 Image Classification ..................................................................... 168
9.3.3.1 Spectral Angle Mapper ................................................. 168 9.3.3.2 Mixture Tuned Matched Filtering .............................. 169 9.3.3.3 Object-Oriented Nearest Neighbor ............................. 170
9.3.4 Accuracy Assessment................................................................... 170 9.4 Results and Discussion ............................................................................ 171 9.5 Conclusion and Future Directions ......................................................... 174 Acknowledgments ............................................................................................. 175 References ........................................................................................................... 176
Accurate and up-to-date information regarding road location and condition is essential data for transportation infrastructure planning and management. This information is used for a variety of purposes including traffic safety, construction projects, traffic engineering studies, and evaluation of maintenance needs [1,2]. Additionally, spatially accurate and up-to-date transportation networks are vital in ambulance and rescue dispatch emergency situations [3,4]. However, current road infrastructure databases are often outdated due to the dynamic nature of road networks where roads deteriorate or are improved and reconstructed [4,5]. In other cases, infrastructure
databases may not even exist at all. This is particularly true for areas experiencing rapid road network expansion [3].