• 论文与研究报告 •

### 基于切平面投影的树干三维表面重建算法

1. 1. 信阳师范学院计算机与信息技术学院 信阳 464000;
2. 中国林业科学研究院资源信息研究所 北京 100091;
3. 信阳师范学院数学与信息科学学院 信阳 464000
• 收稿日期:2016-02-02 修回日期:2016-07-06 出版日期:2016-11-25 发布日期:2016-12-16
• 通讯作者: 唐守正
• 基金资助:

### An Algorithm of Stem Surface Reconstruction Based on Tangent Plane Projection

You Lei1,2, Tang Shouzheng2, Song Xinyu3

1. 1. College of Computer and Information Technology, Xinyang Normal University Xinyang 464000;
2. Research Institute of Forest Resource and Information Techniques, CAF Beijing 100091;
3. College of Mathematics and Information Science, Xinyang Normal University Xinyang 464000
• Received:2016-02-02 Revised:2016-07-06 Online:2016-11-25 Published:2016-12-16

Abstract: [Objective] Three dimension(3D) tree modelling and tree parameters extraction from the three dimensional laser scanning data are the two research focuses in recent years. Combining the two contents, the objective of this study is to reconstruct the irregular triangulation surface of the stem that can be used for extracting stem parameters. [Method] The technology of three dimensional surface reconstruction was used in this study. The algorithm of surface reconstruction which based on tangent plane projection in the three dimensional spaces was improved according to the characteristics of stem point cloud. And the improved algorithm was applied to reconstruct the irregular triangulation as the stem surface model. The improvement was as follows: 1)Radius value was chosen as the criterion for choosing nearest neighbor point set to reduce the influence of the scattered distribution of stem points; 2)The closer the distance between points, the greater the impact between points, the tangent normal vector of a point was calculated by the weighted distance method. And the farther neighbor point was deleted when there are the same project points in the tangent plane of the neighbor point set; 3)According to the geometric topological invariance of the project point in the parallel plane, the calculation of tangent plane was simplified; 4)The transforming of planar point set in the three dimensional spaces to planar point set in the two dimensional spaces was simplified by point set rotating. In the improved algorithm, the current point and its' neighbor point set were projected to the tangent plane and a planar point set was generated. The connected relationship in the Delaunay triangulation of the planar point set was mapped into the stem point. Then, the surface model of the current point and its neighbor points was reconstructed. The global stem surface was obtained through the above reconstruction procedures one point by one point. [Result] The reconstruction experiment of Populus sp. stem showed that the improved algorithm had a better stem model than that of the classic algorithm. The surface reconstruction experiment of stems from different species with different bark properties showed that the reconstructed surface model could clearly show the color information labeled in the field scanning, and the local triangular patch could reveal the scraggly feature of stem surface, and the local triangular patch could show the orientation of the surface triangles. The experiment of extracting stem diameter from the reconstructed surface model showed that, compared with the measured stem diameter in field working, the RMSE value of stem diameter extracted from the reconstructed surface was 0.18 cm, and the accuracy of the reconstructed stem surface was demonstrated.[Conclusion] From the perspective of parameter extraction and using the technology of surface reconstruction, the surface reconstruction algorithm which based on tangent plane projection was improved to reconstructed the stem surface. The reconstructed stem surface had a better visualization effect; it can reflect the real surface features of the tree stem. The improved algorithm can be used for exhibiting the realistic morphological and structural features and constructing realistic three dimensional visualization model of the stem surface. The precise stem reconstructed surface model by the improved algorithm can provide the basic data for stem parameter extraction.