The objective of the research was to investigate by Monte Carlo simulations the effect of tree level errors originating in airborne laser scanning (ALS) on forest planning tree level simulation results. The investigated features were error in the timing of loggings (years) and relative change in net present value. Timing of logging was scrutinized in respect to the next thinning or clear-cut logging simulated for the compartment. Change in a compartment's net present value was calculated for the whole rotation period. The effect of measurement errors in two main tree characteristics, namely tree diameter at breast height and height, on timing of loggings and net present value was studied. The accuracy factors of ALS individual tree detection were based on the EUROSDR/ISPRS Tree Extraction Project in which the quality, accuracy and feasibility of automatic or semiautomatic tree-extraction methods based on high-density ALS data and digital image data were evaluated by a global science network. Research results indicated that errors originating in tree level laser scanning clearly effect simulation results for both the timing and net present value of various types of loggings. The effect of the errors was clearly different in various tree species and development class strata. It was noted that especially errors in tree diameter at breast height measurement effect simulated timing of loggings. Tree height measurement accuracy does not lead to similar consequences. This is a significant observation in respect to the utilization possibilities of tree level ALS as input data for forest planning simulations.


Jensen, K. and Juhl, J.
Gait Analysis by Multi Video Sequence Analysis (pp. 25-34)

The project presented in this article aims to develop software so that close-range photogrammetry with sufficient accuracy can be used to point out the most frequent foot mal positions and monitor the effect of the traditional treatment. The project is carried out as a cooperation between the Orthopaedic Surgery in Northern Jutland and the Laboratory for Geoinformatics, Aalborg University. The superior requirements on the system are that it shall be without heavy expenses, be easy to install and easy to operate. A first version of the system is designed to measure the navicula height and the calcaneus angle during gait. In the introductory phase of the project the task has been to select, purchase and draw up hardware, select and purchase software concerning video streaming and to develop special software concerning automated registration of the position of the foot during gait by Multi Video Sequence Analysis (MVSA). Results show that the developed MVSA system, in the following called Fodex, can measure the navicula height with a precision of 0.5-0.8 mm. The calcaneus angle can be measured with a precision of 0.8-1.5 degrees.


Jokinen, O.

The paper deals with digitizing the 3-D shape of a trunk from a sequence of image pairs when a harvester is approaching the tree. The stereo correspondence problem is solved hierarchically using natural features, image correlation, and epipolar constraints. False matches are removed by analyzing the distribution of reconstructed 3-D points and applying a novel similarity criterion. Two methods for solving the motion between successive frames are considered. The first one is based on tracking the stereo correspondences to the next frame using image correlation. The second one is based on registering 3-D point sets of successive frames using surface matching techniques. In the first case, the final surface model is triangulated from points successfully tracked through several frames while in the second case, it is triangulated from points where the correspondences are compatible in surface matching. Test results verify the performance of the methods in a pine forest.


Vastaranta, M., Melkas, T., Holopainen, M., Kaartinen, H., Hyyppä, J., Hyyppä, H.
Laser-Based Field Measurements in Tree-Level Forest Data Acquisition (pp. 51-61)

Remote-sensing (RS) methods, such as airborne laser scanning (ALS), are capable of acquiring tree-level forest information with high accuracy. However, these methods need more detailed reference data from sample plots than standwise mean characteristics. Field-measured treewise characteristics must be matched with ALS data on the measured position of the tree. In addition to tree position and species, diameter-at-breast height (dbh) measurements are an essential part of these reference data, since dbh at the individual tree level cannot be measured directly from the ALS data. Here we compare the accuracy in treewise field measurements with different laserbased methods in practice. The characteristics under observation include dbh, tree height and stem position. The comparable instruments used include the Terrestrial laser scanner (TLS), laser-relascope and laser-camera. The data consist of 122 trees from sample plots in Saunalahti and Nuuksio. The standard errors in dbh measurements were 8.3 mm (4.5%), 8.5 mm (4.9%) and 14.3 mm (8.3%) with the TLS, laser-camera and laser-relascope, respectively. The standard errors in positional accuracy of the tree were 0.1 m with the TLS and 0.7 m with the laserrelascope. A bias of -0.6 m (-5.2%) and standard error of 1.0 m (10.1%) were present in tree height measurements with the laser-relascope. The results of the study can be utilized e.g. in developing new cost-efficient methods for collecting ground truth data for ALS-based inventories.