Rapid Extraction and Update of Road Network to Caltrans Database

2 DISCLAIMER The contents of this report reflect the views of the authors, who are responsible for the facts and the accuracy of the information presented herein. Department of Transportation in the interest of information exchange. The U.S. Government and California Department of Transportation assume no liability for the contents or use thereof. The contents do not necessarily reflect the official views or policies of the State of California or the Department of Transportation. ABSTRACT The goal of this research is to develop new approach and technique to improve and extend the capabilities of creating, modeling and maintaining accurate and up-to-date road infrastructure databases for transportation managements and services. Our research efforts are to assess, define, and use the unique spatial and spectral characteristics of the new, advanced sensor techniques from aerial imagery and LiDAR for automated road extraction and road quality mapping. A number of theoretical and experimental studies lead us to pursue an innovative approach that merges the power of perceptual grouping with sensor cues, geometric invariants, and machine learning under a unified framework to tackle these problems. This new approach has the potential for automating the extraction and mapping of complex road networks from remote sensing data. In addition, the same process also allows for a constrained optimal estimation of various terrain features and attributes, thereby producing hierarchical data representations under a consistent framework. Most important, we believe that the process of labeling the model elements as buildings, vegetation, roads, and terrains will be possible within this framework. We anticipate a significant step reduction in the human time and effort required to produce and update accurate road models to transportation infrastructure databases.