This work reviewed various tree detection and delineation algorithms and developed another one for ESRI ArcGIS Desktop software. This algorithm which employs the principles of local maxima filtering and region growing technique was enhanced by successfully applying NDVI index. Consequently, this algorithm was tested on 11 sample plots. These plots were located in the forest stand 187 C 10a (according to Forest Management Plan 2013 - 2022) in the forest section Proklest which belongs to the forest district Habrůvka in The Training Forest Enterprise Masaryk Forest Křtiny. The results were consequently, evaluated using these 6 different spatial resolutions: 10 cm, 20 cm, 40 cm, 60 cm, 80 cm and 100 cm. This evaluation was based on the comparison with manually vectorised 1,560 tree crowns. Spatial resolution of 40 cm provided the most accurate results for tree counting. At a standard deviation of 7.8 %, the accuracy reached 98 % of the reference layer. In the case of tree crown delineation, the most accurate results were provided for the spatial resolution of 80 cm. The accuracy reached 103 % with standard deviation of 9.6 %.
| Identifer | oai:union.ndltd.org:nusl.cz/oai:invenio.nusl.cz:190595 |
| Date | January 2015 |
| Creators | Miklas, Vojtěch |
| Source Sets | Czech ETDs |
| Language | English |
| Detected Language | English |
| Type | info:eu-repo/semantics/masterThesis |
| Rights | info:eu-repo/semantics/restrictedAccess |
Page generated in 0.0019 seconds