Automatic tree detection in sample plots from a simple terrestrial laser scanning
This study aimed the automatic identification of trees in circular sample units of 200, 300 and 400 m² in stands of Eucalyptus spp., using TLS (Terrestrial Laser Scanning) in Mato Grosso do Sul State, Brazil. Three planting ages were tested: 2 years-old-plantation, considering trees with natural pruning (E. urophylla -200 m²) and without natural pruning (E. grandis - 300 m²), and trees with 4.5 and with 5.5 years-old without pruning (hibrid E. urophylla x E. grandis - 400 m²). Field laser scanning was performed with Trimble TX5 equipment. This approach was based on a cut of point-cloud at 1.3 m above ground level (DBH) and the developing the algorithm 2D slice. It was possible to detect 98.3, 98 and 93.9% of trees for samples units of 200, 300 and 400 m², respectively. These identifications were influenced by the spatial distribution of trees because of the shading of trees further from the equipment. This simple scan should be applied only for small sample units. The error detections in areas greater than 200 m², may influence negatively the dendrometric estimations.
| Main Authors: | , , , , |
|---|---|
| Format: | Digital revista |
| Language: | por |
| Published: |
Embrapa Florestas
2019
|
| Online Access: | https://pfb.cnpf.embrapa.br/pfb/index.php/pfb/article/view/1730 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This study aimed the automatic identification of trees in circular sample units of 200, 300 and 400 m² in stands of Eucalyptus spp., using TLS (Terrestrial Laser Scanning) in Mato Grosso do Sul State, Brazil. Three planting ages were tested: 2 years-old-plantation, considering trees with natural pruning (E. urophylla -200 m²) and without natural pruning (E. grandis - 300 m²), and trees with 4.5 and with 5.5 years-old without pruning (hibrid E. urophylla x E. grandis - 400 m²). Field laser scanning was performed with Trimble TX5 equipment. This approach was based on a cut of point-cloud at 1.3 m above ground level (DBH) and the developing the algorithm 2D slice. It was possible to detect 98.3, 98 and 93.9% of trees for samples units of 200, 300 and 400 m², respectively. These identifications were influenced by the spatial distribution of trees because of the shading of trees further from the equipment. This simple scan should be applied only for small sample units. The error detections in areas greater than 200 m², may influence negatively the dendrometric estimations. |
|---|