ABSTRACT
ABSTRACT: We discuss the feasibility of using airborne LiDAR imagery data to support traffic flow parameter estimation, including vehicle count estimates and vehicle classification, and to a lesser extent, velocity estimates. As a part of the classification task, we demonstrate the capability of LiDAR data to efficiently identify vehicles or vehicle categories by shape. We show that LiDAR offers the capability to better preserve the vehicle geometry, especially the vertical profile, as compared to optical imagery. During the projection process in optical imagery the vertical dimension is usually lost. Thus, a better vehicle classification/grouping was found from using dense LiDAR data. The identified and categorized vehicles can directly support the vehicle count estimation process. Experimental results are presented to validate the performance potential of the LiDAR data-based vehicle extraction process.
