While the process of data collection, as well as the sensors available are rapidly improving, processing and managing the data generated can remain a challenge. This session showcases advances in lidar processing in real-world projects. Hear from the professionals responsible for going from the field to providing the finished deliverable – and how they got there.
Cleaning and Applications of Shoreline Point Cloud Data
This presentation will focus on multiple related projects that revolve around cleaning and finding new applications for a massive dataset of shoreline lidar data that covers over 280 miles of the Lower Mississippi River. A new virtual reality based point cloud editing tool was developed, and experimentally shown to be significantly faster for cleaning point cloud data than traditional desktop tools. The 3D VR interface matches the dimensionality of the data, providing superior perception of point data, and increased efficiency of editing interaction. Geospatial references are provided to assist the hard-to-automate subjective decisions that must often be made during cleaning. This VR tool has been released as free software that works with most PC-based VR systems and can be used to edit any type of point cloud data.
After being cleaned, the Mississippi River shoreline lidar data is incredibly useful. It has been incorporated into a web-based 3D mapping and journey planning interface, where it provides detailed representations of harbor infrastructure. The data is also used to calculate highly-accurate 3D clearance bounds for bridges and other overhead obstacles, which are used to provide mariners with real time air gap clearance information, critical to enabling precision navigation of ever-larger cargo vessels.
Thomas Butkiewicz, University of New Hampshire
Machine Learning for Classifying Lidar Point Clouds
The State of Texas connected with Fugro to enhance lidar point cloud classifications of USGS QL2 lidar data using machine learning techniques. The projects covered over 100,000 square miles of South, Central, and East Texas. Fugro added low, medium, and high vegetation, buildings, and culverts to the USGS lidar data. Machine learning techniques allowed Fugro to classify the clusters of points to a high degree of accuracy. This session will showcase project details, data management, results of the newly classified lidar point cloud, a comparison to traditional techniques and accuracy. We’ll close out with examples of lidar data use-cases as a result of the better classified lidar data.
Keith Owens, Fugro