| The tree should be felled to measure diameter at any heightin tradition. This traditionalmeasurement is destructive and time-consuming. Under the theme of ecological construction and the prohibition of felling wildwood, a nondestructive measuring mode should be developed for standing tree measurement. Terrestrial laser scanner(TLS), with millimeter accuracy of measurement, has the capability of obtaining large-scale point cloud from the surface in a short time. Itcanprovide an effective way for measuring diameter at any height of the stem. In recent years, extracting forestry parameters and constructing tree models from TLShave become research hotspots. However, according to the existing studies, there are some problem existing in forestry measurement and tree modeling.(1)The stem point cloud used for extracting the stem diameter did not established on the basis of the stem cross-section.(2)The shape of the stem cross-section is an irregular geometry; the diameter value extracted by circle fitting method is not accurate.(3)The extracted parameters didn’t including stem tilt, curvature and torsion.(4)The stem skeleton represented by piecewise polygonal line is difficult for scientific calculation.(5)The reconstructed stem surface model could not reflect the scraggly characteristic of bark. That determines the reconstructed model cannot use for extracting stem diameter, stem cross-sectional area and other stem parameters. For the above problems, the aim of this study is to build a set of accurate methods for extracting stem parameters from stem point cloud.In this study, the research species including Populus sp., Fraxinus pennsylvanica marsh., Fraxinus rhynchophyllavar. rhynchophylla Hemsl., Platanus occidentalis L., Ailanthus altissima(Mill.) Swingle., Larix gmelinii(Rupr.) Rupr. The trees were scanned by TLS and measured with caliper and diameter tape in the field work. On the basis of cloud point preprocessing, the following research content were carried out in this study.(1)An algorithm of extracting stem point based on tangent plane was present.(2)An algorithm of building three dimensional spacesstem axis curvethat can be used for accurate measurement was present.(3)Two methods of simulating diameter tape measurement and a method of simulating caliper measurement were present.(4)A method of constructing stem cross-sectional profile curve and accurate calculating stem cross-sectional area was present.(5)A fast growth algorithm centered on priority point for Delaunay triangulation was present. The surface reconstruction algorithm based on tangent plane projection was improved and applied to the stem surface reconstruction. An stem surface reconstructionalgorithm based on cylindrical projection was present. A quantitative evaluation method for surface reconstruction model was present.Through the above mentioned research, the following conclusions were reached.(1)The stem point extracting algorithm was suit for extracting stem point from tree with little tilt and bending.(2)The three dimensionalspace stem axis curve canreflectthe geometrical characteristicof the stem. The stem cross-section at any heightcan be located accurately. The curvature, torsion and other parameters also can be calculated accurately.(3)Compared with the diameter in the field work, the RMSE value of extracting diameter by simulating track of diameter tape using a cubic B-Spline and a piecewise Bezier curve were 0.16 cm and 0.15 cm, respectively. The two methods of extracting diameter by simulating track of diameter tape were more accurate thanthe other existing methods. The method of piecewise Bezier curve was the best one. Compared with diameter of caliper in the field work, the RMSE value of extracting diameter by simulating caliper was 0.34 cm.(4)Themethod ofstem cross-sectional area calculation of this studyis more accurately than the traditional circle area method. Compared with the method of this study, the MAPE value of the stem cross-sectional area by the circle area method was 6.20%.(5)The time of constructing Delaunay Triangulation didn’t exceed one third time of the original classical algorithm. Compared with the Powercrust surface reconstruction algorithm and the surface reconstruction algorithm based on isosurface proposed by Hoppe, the stem surface reconstruction algorithms of this study can better reflect the scraggly characteristics of the stem bark. The surface reconstruction algorithm based on cylindrical projection had the best result. According to the quantitative evaluation data of extracting diameter from reconstructed surface, compared with the diameter in the field work, the RMSE value of the extracting diameter from surface reconstructed based on tangent plane projection and cylindrical projection were 0.14 cm and 0.15 cm. Compared with the diameter extracted from stem point cloud, the RMSE value of extracting diameter from surface reconstructed based on tangent plane projection and cylindrical projection were 0.11 cm and 0.05 cm. The surface reconstructed algorithm based on cylindrical projection was suit for stem surface reconstruction. |
PDF Full Text Request