We present a highly-accurate tree species recognition approach that uses only 3D spatial terrestrial laser scanning data (TLS) and is fully automatic. Individual trees are automatically separated from forest plot level point clouds and reconstructed as a cylindrical quantitative structure model (QSM). As a QSM contains the comprehensive geometric and topological structure of a tree, we have access to tree properties that are impossible to compute directly from point cloud data without reconstructing full QSMs. In comparison, previous methods require at least some level of manual interaction to separate individual trees, and they often rely on additional data sources, e.g., hyperspectral data.

We defined 15 classification features and tested their performance on over 1200 trees from 3 different species, and 5 different classification methods. We used trees from three single-species and two mixed-species forest plots, all located in Finland. The results show that accurate species classification based on TLS data and QSMs is possible when suitable training data are available. Our tests showed that as little as 30 training samples per species can produce good results. All of the classification methods performed well and maximum classification accuracy was over 95 % for all of the methods.

For about two years our research team at Tampere University of Technology has been disseminating research results in the form of animations, interactive 3D models and Facebook updates. The animations showcase how the methods we have developed can be used to reconstruct tree models from terrestrial laser scanning data. The computations are done in Matlab, and the results are exported to Blender - an open-source 3D modeling and animation software - to make data-driven animations that are later uploaded to the group's Youtube channel and further shared on our research consortium's Facebook page. Some of the resulting models have also been published as interactive 3D models that can be viewed in a standard web browser supporting WebGL technology. The extensive, positive feedback has shown that these innovative dissemination channels are vital for helping others understand the research methods and results without getting lost in the details.

A poster (in Finnish) about how new technologies will change forest management over the next few years. More detailed forest information increases the value of individual trees. The poster was showcased in the TTY Forum 2014 event in Tampere-talo 4th October, 2014.

The purpose of this study was to advance numerical methods for radio tomography in which asteroid's internal electric permittivity distribution is to be recovered from radio frequency data gathered by an orbiter. The tomography approach examined was closely related to that of the CONSERT experiment aiming at recovery of a comet nucleus structure as a part of the ROSETTA mission. The focus was on signal generation via multiple sources (transponders) providing one potential, or even essential, scenario to be implemented in a challenging in situ measurement environment and within tight payload limits. The permittivity was reconstructed with a combination of the iterative alternating sequential (IAS) inverse algorithm and finite-difference time-domain (FDTD) forward simulation. Single and multiple source scenarios were compared in two-dimensional localization of permittivity anomalies.

We present a computational method that produces automatically precision models of trees from terrestrial laser scanning (TLS) data. The method is fast, typically few minutes per tree, and the resulting model contains both the topological and geometrical information of the tree. The method is validated using artificial and real TLS data. The results show that TLS together with computational reconstruction method provides means to collect structural information of trees fast and nondestructively.

We present comprehensive and quantitative tree models reconstructed from terrestrial laser scanning data. The tree models consist of large number of cylinders whose location, size, and orientation locally approximate the geometry of the tree. The parent-child relations of the cylinders also record the topological branching structure of the tree. The modeling process is automatic and scale-independent. When the tree model is computed once, it can be used to compute tree attributes, such as branch size distributions and taper functions, without the need to revisit the original dataset. The model is also compact, achieving a hundred- to thousandfold data size compression compared to the original dataset. We present also a validation of the model using generated tree models and examples of models from measurements of actual trees.