Classified Point Cloud

1. LiDAR Calibration/Automatic Classification. After initial data post processing, point cloud data processing included calibration to project ground control and initial automated classification of point cloud data. LiDAR data were inventoried and tested for preliminary vertical accuracy verification against project control. Point cloud source data were processed initially as ground (class 2) or non-ground points (class 1) through automated means in TerraScan® software.

2. Production Staging. Analysis of point cloud data was conducted to produce full production tile sets. This inventory step ensured that overlapping production tiles were considered to provide for a consistent production tiling scheme without overlap. Once a production tiling scheme was organized, a full set of quality control (QC) shaded relief products were derived from the initial automatic classification using Geospatial Data Abstraction Library (GDAL) Utilities. Data processing priorities were established to initial focus on urbanized areas.

3. Review and Classification. Shaded relief QC products were distributed to analysts for review. Analysts placed polygonal calls identifying anomalies categorized into building, bridge, pit/spike, mine, ridge, and other. Building calls drove manual reclassification of points from ground to unclassified using ERDAS Imagine and Safe Software FME. Bridge calls drove establishment of breakline features and subsequent manual review to ensure correct classification of all ground points around breaklines and bridge decks. Pit/Spike calls were addressed through automatically reclassification by statistical means which explored for data outliers falling outside of three (3) standard deviations from the mean surface. Polygonal ridge calls were used as a data clipper to clip out relevant sections of the point cloud for reclassification of the section using ERDAS Imagine. Once the ridge segment was reclassified in ERDAS it was merged back into the original production classified point cloud. A feedback loop at the end of the review and classification process provided an updated shaded relief image. A second analyst verified all calls to ensure correct classification.

4. Breakline Classification. Sparse breakline data was integrated for use in classification of point cloud data. Breaklines were used as reference to classify water (9), Ignored ground (near a breakline) (10) and bridge decks (17). For waterbodies, extracted breaklines were conditioned to the LiDAR elevation using a review of contours and shaded relief imagery.