文章基本信息

标题：Approaches for Recognition of Wood Quality of Standing Trees Based on Terrestrial Laser-Scanner Data
本地全文：下载
作者：C. Schütt ; T. Aschoff ; D. Winterhalder 等
期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2004
卷号：XXXVI-8/W2
出版社：Copernicus Publications
摘要：In this project the applications of terrestrial laserscanner for recording characteristics of quality of standing trees are examined. Furthermore it is investigated to what extent inhomogeneous bark structures, for example, branch scar, blind conk etc. can be recognized and so indicate the inner wood quality of a tree. The effects of the outward wood defects are explained by means of stems which were finished to sliced veneer. The interesting parameters are, for example, the expected block-, or rather, the log- length, the ovality in different heights, the volume of the log as well brachiness, numbers and shape of the branch scars. This method of measurement hopes to contribute to the sustainability in value and enable approximate through-cut calculations of standing trees. This problem was exemplarily investigated by analyzing seven logs of the tree species European beech (Fagus sylvatica), oak (Quercus spec.) and wild cherry (Prunus avium). To detect the outward wood defects on the scanned trees the laserscans were unwound and analyzed. Firstly the trees were identified in the laserscanner data, and for every trunk, the diameters were calculated of the log ends. These diameters were averaged and yielded the diameter of the reference cylinder, which constitutes an ideal trunk. The scan coordinates were converted into picture coordinates by means of different transformation matrices. The result of this cylindrical unwinding is a 2D-picture for every scanned tree. These pictures were applied to another analysis software, which enabled an semiautomatic wood defect detection and classification. The software works with different image algorithms as well as with a neural network, so that the recognition rate of the software can be improved by reading of training data records, and furthermore, learn from each additional picture analysis. The result of the analysis was the corresponding classification of every wood defect as well as its location and expansion on the tree stems. Through the combination of the results a "decision tree" is created, which later should support the cut calculation. For example, it is possible, that in the " decision tree" define how many branch scars of a specific diameter inside a specific log can be found for a defined quality, and that the result of the cut calculation's software provides the corresponding dimension and the expected log volume
关键词：laserscans; intensity; range; unwind; wood defects; wood quality