Share Email Print

Proceedings Paper

Terrestrial laser scanning for biomass assessment and tree reconstruction: improved processing efficiency
Author(s): Ahmed Alboabidallah; John Martin; Samantha Lavender; Victor Abbott
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Terrestrial Laser Scanning (TLS) processing for biomass mapping involves large data volumes, and often includes relatively slow 3D object fitting steps that increase the processing time. This study aimed to test new features that can speed up the overall processing time. A new type of 3D voxel is used, where the horizontal layers are parallel to the Digital Terrain Model. This voxel type allows procedures to extract tree diameters using just one layer, but still gives direct tree-height estimations. Layer intersection is used to emphasize the trunks as upright standing objects, which are detected in the spatially segmented intersection of the breast-height voxels and then extended upwards and downwards. The diameters were calculated by fitting elliptical cylinders to the laser points in the detected trunk segments. Non-trunk segments, used in sub-tree- structures, were found using the parent-child relationships between successive layers. The branches were reconstructed by skeletonizing each sub-tree branch, and the biomass was distributed statistically amongst the weighted skeletons. The procedure was applied to nine plots within the UK. The average correlation coefficients between reconstructed and directly measured tree diameters, heights and branches were R2 = 0.92, 0.97 and 0.59 compared to 0.91, 0.95, and 0.63 when cylindrical fitting was used. The average time to apply the method reduced from 5hrs:18mins per plot, for the conventional methods, to 2hrs:24mins when the same hardware and software libraries were used with the 3D voxels. These results indicate that this 3D voxel method can produce, much more quickly, results of a similar accuracy that would improve efficiency if applied to projects with large volume TLS datasets.

Paper Details

Date Published: 6 September 2017
PDF: 10 pages
Proc. SPIE 10444, Fifth International Conference on Remote Sensing and Geoinformation of the Environment (RSCy2017), 104440C (6 September 2017); doi: 10.1117/12.2279130
Show Author Affiliations
Ahmed Alboabidallah, Plymouth Univ. (United Kingdom)
John Martin, Plymouth Univ. (United Kingdom)
Samantha Lavender, Pixalytics Ltd. (United Kingdom)
Victor Abbott, Plymouth Univ. (United Kingdom)

Published in SPIE Proceedings Vol. 10444:
Fifth International Conference on Remote Sensing and Geoinformation of the Environment (RSCy2017)
Kyriacos Themistocleous; Silas Michaelides; Giorgos Papadavid; Vincent Ambrosia; Gunter Schreier; Diofantos G. Hadjimitsis, Editor(s)

© SPIE. Terms of Use
Back to Top