Home About Browse Search
Svenska


Axelsson, Arvid, 2018. Using multispectral ALS for tree species identification. Second cycle, A2E. Umeå: SLU, Dept. of Forest Resource Management

[img]
Preview
PDF
1MB

Abstract

Accurate and large area tree species classification is an important subject with problems that have not yet been completely solved. For both nature conservation and wood production purposes, a detailed description of tree species composition would be useful. The objective of this master’s thesis is to explore how tree species differ in spectral and structural properties using multispectral airborne laser scanning data from the Optech Titan X system. Remote sensing data was gathered from Remningstorp, Västra Götaland in Sweden on 21st July 2016. Field data contained 179 solitary trees from nine species. Two new methods for feature extraction are tested and compared to features of height and intensity distributions. The features that were most important for tree species classification were those from the upper part of the crown. Spectral features provided a better basis for tree species classification than structural features. Using single, first or all returns gave only a small difference in cross-validation correctness rate. The best classification model was created using multispectral distribution features of all returns, with an correctness rate of 77.09 %. Spruce and pine had a 100 % overall classification accuracy and were not confused with any other species. Linden was the deciduous species with a large sample that was most frequently confused with many other deciduous species.

Main title:Using multispectral ALS for tree species identification
Authors:Axelsson, Arvid
Supervisor:Lindberg, Eva
Examiner:Olsson, Håkan
Series:Arbetsrapport / Sveriges lantbruksuniversitet, Institutionen för skoglig resurshushållning och geomatik
Volume/Sequential designation:491
Year of Publication:2018
Level and depth descriptor:Second cycle, A2E
Student's programme affiliation:SY001 Forest Science - Master's Programme 300 HEC
Supervising department:(S) > Dept. of Forest Resource Management
Keywords:LiDAR, indvidual trees, ITC, spectral
URN:NBN:urn:nbn:se:slu:epsilon-s-10194
Permanent URL:
http://urn.kb.se/resolve?urn=urn:nbn:se:slu:epsilon-s-10194
Subjects:Forestry - General aspects
Forestry production
Language:English
Deposited On:13 Feb 2019 12:42
Metadata Last Modified:14 Feb 2019 02:00

Repository Staff Only: item control page

Downloads

Downloads per year (since September 2012)

View more statistics

Downloads
Hits