Use of Time-Stack Processing in Vehicle Tracking from Roadside 3D LiDAR
MetadataShow full item record
Publisher:The Ohio State University
Series/Report no.:The Ohio State University. Department of Electrical and Computer Engineering Honors Theses; 2020
Light Detection and Ranging (LiDAR) is a perception technology that is widely used in Intelligent Transportation System (ITS) applications. This thesis presents a time-stack based method for tracking vehicles through 3D LiDAR point clouds recorded from a LiDAR scanner installed next to a roadway. The time-stack methodology is adapted from video image processing techniques that sample a 1D or 2D region at fixed time intervals to compile a 2D or 3D "time-stack". The time-stack is a high dimension representation of spatiotemporal data with changing features. For a 1D region, the time-stack is essentially a time-space diagram showing how the visual features evolved through the region over time and space. Various techniques can then be used to find and trace the features in the time-stack in a model-free tracking methodology. In this thesis the time-stack is modified from the conventional video-based format to instead present the evolution of vehicle locations as they travel through the LiDAR field of view. The scope of this work spans from data preparation to post-processing. The data preparation ultimately identified confounding factors in the data collection, and developed techniques to address these problems. The research developed techniques for ground identification, road boundary identification, vehicle clustering, and vehicle tracking. For a given data scan, the static background was identified and discarded, any remaining data points within the road boundaries were clustered to form vehicles. A linear road model was introduced as a referencing system for measuring distance traveled along the roads for any vehicle. The processing techniques developed from this research are anticipated to contribute to the development of traffic surveillance and offer valuable insights into driving behaviors.
Academic Major: Electrical and Computer Engineering
National Science Foundation