Automatic Recognition of Diseased Trees Based on the Vertical Structure of Airborne Point Clouds: A Case Study of Diseased Trees of Great Smoky Mountains
Hu W, Wang Y, Hu Q. 2016. Automatic Recognition of Diseased Trees Based on the Vertical Structure of Airborne Point Clouds: A Case Study of Diseased Trees of Great Smoky Mountains. In 2016 Fourth International Workshop on Earth Observation and Remote Sensing Applications
ABSTRACT: This paper introduces a method to recognize the area of diseased trees from LiDAR point clouds automatically. Airborne LiDAR technology can acquire high-precision 3D information of tree structures, and provides a new possibility to monitor forest diseases and insects on a large scale. Based on the differences of tree canopy of the diseased and healthy trees, this paper researches the vertical structure feature lines of trees, and proposes an approach of automatic diseased tree recognition. Firstly, the point clouds are filtered into ground points and nonground points. Then the normalization process computes nDSM (normalized digital surface model) which replaces the raw elevation with relative heights of the non-ground points. Finally, non-ground points with local height values are divided into identical grids in the projection horizontal plane. The vertical structure features of the points in each layer of each grid are calculated and analyzed to get feature lines, which is used to identify the diseased trees by comparing with the reference feature lines. The experiment data is the LiDAR point clouds of Great Smoky Mountains, where the living situation of local trees is under the invasion of non-native insects and fungi over the years. The experimental results show that this approach has superiority and effectiveness in automatic recognition of the disease trees.
- Type
- Conference Proceeding Paper
- Authors
- Hu, Wei; Wang, Yue; Hu, Qingwu
- Units
- GRSM
- Keywords
- APHN, Appalachian Highlands Network, automatic recognition, diseased trees, Great Smoky Mountains National Park, GRSM, LiDAR, point cloud normalization, SER, Southeast Region, vertical structure line