中文摘要：

[目的]树干在林业生产与科研中具有重要的地位,树干表面重建对森林可视化与树干参数提取具有重要意义,本研究以重建树干的精确表面模型为目的。[方法]提出了一种基于圆柱面投影的树干表面重建算法,算法以连续的多个树干分段为重建单元,以重建单元的质心与直径构建圆柱面,通过将树干点云投影至圆柱面,再将圆柱面点云延伸展开得到平面点集,由平面点集Delaunay三角网的连接关系构建重建单元的不规则三角网表面,连接相邻重建单元的树干表面以获取树干整体的表面模型。[结果]在树皮粗糙程度各异的3种树干点云上的实验表明：重建的树干表面能清晰地展示外业时标注的信息,能有效地反映树干表面的褶皱凸凹特征,具有更好的可视化效果;通过从重建表面上定量提取直径的精确性评价表明：与围尺实测直径相比,从本文提出的基于圆柱面投影重建的树干表面上提取树干直径的RMSE为0.14 cm,比基于切平面投影重建方法更精确。[结论]本文提出的算法是一种简单高效的树干表面重建算法,无需计算法向量、重建的树干表面上三角形数量较少,且能有效地还原树干表面的凸凹特征,适用于树干表面的精确重建、精确可视化与树干参数提取。

英文摘要：

[Objective]To reconstruct the tree stem surface model. [Methods]A three-dimensional stem surface reconstruction algorithm based on cylindrical projection was presented. A number of consecutive slices were viewed as the reconstruction units. A cylinder was constructed by the centroid and diameter of the reconstruction unit. The stem points of reconstruction unit were transformed into cylindrical points,and then the cylindrical points were unfolded extendedly to planar points. The surface of reconstruction unit was reconstructed according to the connection among points in Delaunay triangulation created from planar point. The whole stem surface was obtained by connecting reconstructed surface of adjacent reconstruction unit. [Results]The experiment based on the stem point set of the barks of three tree species with different roughness showed that the reconstructed surface by proposed algorithm can showed information labeled in field work clearly and can exhibit the roughness characteristic of stem bark. By comparing with other three surface reconstruction algorithms,the proposed algorithm showed better visualizationeffect. Through quantitative analysis of diameter extracting from reconstructed surface,and comparing with the diameter by tape diameter in field work,the RMSE value of diameter extracting surface reconstructed by proposed algorithm was 0. 14 cm and the surface reconstructed by proposed algorithm was more accurate than the algorithm of tangent based. [Conclusion]The present method is a simple,efficient and accurate one for stem surface reconstruction without calculating the normal vector and needs fewer surface triangles. It can effectively exhibit the surface features of the stem. The proposed algorithm was suitable for surface reconstruction,exact visualization of the stem and extracting stem parameters in an exact way.