Public defence in geoinformatics, M.Sc. Jussi Juola
When
Where
Event language(s)
New hyperspectral technology shows promising results for automatic identification of tree species
A new camera technology can be used to capture how tree stems interact with light in unprecedented detail. This information on how stems reflect light is valuable in remote sensing of forests, biodiversity mapping, and forestry applications.
For the first time, this dissertation introduces unique field and laboratory measurement set-ups that were used to capture hyperspectral data of tree stems for boreal and temperate tree species.
The results showed that the stems of different tree species reflect light differently in the visible and near-infrared wavelengths. The distinctive spectral properties and patterns in the bark performed efficiently in the automatic identification of tree species. Using modern computational methods, such as convolutional neural networks, up to 97.5% overall accuracy was achieved in identifying tree species. In addition to identifying tree species, this fine-scale information can help us to understand remotely sensed signals of forests. In the future, close-range hyperspectral imaging could open new possibilities for mapping tree species, for example, in precision-forestry with autonomous wood harvesters or in citizen science with personal smartphones.
Video summary of the dissertation: https://www.youtube.com/watch?v=Mx-3tcKjNxk
Opponent: Dr. Michael Förster, Technical University of Berlin, Germany
Custos: Professor Miina Rautiainen, Aalto University School of Engineering, Department of Built Environment
Contact information of the doctoral student: Jussi Juola, jussi.juola@aalto.fi, mob. +358504999186
The public defence will be organised on campus (auditorium M1, Otakaari 1).
The thesis is publicly displayed 10 days before the defence in the publication archive Aaltodoc of Aalto University.