Browsing by Author "Vaaja, M. T."

Now showing 1 - 4 of 4

Approaches for Mapping Night-Time Road Environment Lighting Conditions
(2020-08-03) Vaaja, M. T.; Maksimainen, M.; Kurkela, Matti; Virtanen, Juho-Pekka; Rantanen, T.; Hyyppä, H.
A4 Artikkeli konferenssijulkaisussa
The integration of the 3D measurement techniques with luminance imaging has increased the potential for mapping night-time road lighting conditions. In this study, we present selected static and mobile approaches for the purpose. The measurement methods include conventional 2D imaging luminance photometry and the integration of the luminance imaging with terrestrial and mobile laser scanning. In addition, we present our initial experiences with performing integrated luminance mapping and photogrammetric reconstruction from drone imagery. All of the presented methods require that the camera is calibrated with a reference luminance source. Our results show the results of luminance calibration and feasibility of 3D luminance point clouds for evaluating road surface luminances. In addition, we discuss the other potential applications, limitations and future research.
THE EFFECT of WIND on TREE STEM PARAMETER ESTIMATION USING TERRESTRIAL LASER SCANNING
(2016-06-07) Vaaja, M. T.; Virtanen, J. P.; Kurkela, M.; Lehtola, V.; Hyyppä, J.; Hyyppä, H.
A4 Artikkeli konferenssijulkaisussa
The 3D measurement technique of terrestrial laser scanning (TLS) in forest inventories has shown great potential for improving the accuracy and efficiency of both individual tree and plot level data collection. However, the effect of wind has been poorly estimated in the error analysis of TLS tree measurements although it causes varying deformations to the trees. In this paper, we evaluated the effect of wind on tree stem parameter estimation at different heights using TLS. The data consists of one measured Scots pine captured from three different scanning directions with two different scanning resolutions, 6.3 mm and 3.1 mm at 10 m. The measurements were conducted under two different wind speeds, approximately 3 m/s and 9 m/s, as recorded by a nearby weather station of the Finnish Meteorological Institute. Our results show that the wind may cause both the underestimation and overestimation of tree diameter when using TLS. The duration of the scanning is found to have an impact for the measured shape of the tree stem under 9 m/s wind conditions. The results also indicate that a 9 m/s wind does not have a significant effect on the stem parameters of the lower part of a tree (<28% of the tree height). However, as the results imply, the wind conditions should be taken into account more comprehensively in analysis of TLS tree measurements, especially if multiple scans from different positions are registered together. In addition, TLS could potentially be applied to indirectly measure wind speed by observing the tree stem movement.
On Selecting Images from An Unaimed Video Stream for Photogrammetric Modelling
(2020-08-03) Rönnholm, P.; Vaaja, M. T.; Kauhanen, H.; Klockars, T.
A4 Artikkeli konferenssijulkaisussa
In this paper, we illustrate how convolutional neural networks and voxel-based processing together with voxel visualizations can be utilized for the selection of unaimed images for a photogrammetric image block. Our research included the detection of an ear from images with a convolutional neural network, computation of image orientations with a structure-from-motion algorithm, visualization of camera locations in a voxel representation to detect the goodness of the imaging geometry, rejection of unnecessary images with an XYZ buffer, the creation of 3D models in two different example cases, and the comparison of resulting 3D models. Two test data sets were taken of an ear with the video recorder of a mobile phone. In the first test case, a special emphasis was taken to ensure good imaging geometry. On the contrary, in the second test case the trajectory was limited to approximately horizontal movement, leading to poor imaging geometry. A convolutional neural network together with an XYZ buffer managed to select a useful set of images for the photogrammetric 3D measuring phase. The voxel representation well illustrated the imaging geometry and has potential for early detection where data is suitable for photogrammetric modelling. The comparison of 3D models revealed that the model from poor imaging geometry was noisy and flattened. The results emphasize the importance of good imaging geometry.
UTILISING SIMULATED TREE DATA TO TRAIN SUPERVISED CLASSIFIERS
(2022-05-30) Rönnholm, P.; Wittke, S.; Ingman, M.; Putkiranta, P.; Kauhanen, H.; Kaartinen, H.; Vaaja, M. T.
A4 Artikkeli konferenssijulkaisussa
The aim of our research was to examine whether simulated forest data can be utilized for training supervised classifiers. We included two classifiers namely the random forest classifier and the novel convolutional neural network classifier that utilizes feature images. We simulated tree parameters and created a feature vector for each tree. The original feature vector was utilised with random forest classifier. However, these feature vectors were also converted into feature images suitable for input into a YOLO (You Only Look Once) convolutional neural network classifier. The selected features were red colour, green colour, near-infrared colour, tree height divided by canopy diameter, and NDVI. The random forest classifier and convolutional neural network classifier performed similarly both with simulated data and field-measured reference data. As a result, both methods were able to identify correctly 97.5 % of the field-measured reference trees. Simulated data allows much larger training data than what could be feasible from field measurements.