Abstract:

Objective: A reconstruction method for the missing parts of tree trunks for stem shape control is proposed to address the issue that when conducting forestry investigations using terrestrial laser scanning (TLS) technology, the upper part point cloud scanning of trees is incomplete due to the occlusion of branches and adjacent trees, resulting in the reduction of the extraction accuracy of parameters such as tree height and diameter. Method: In this study, 138 Larix olgensis trees in different sites and age groups were selected from a L. olgensis plantation in Mengjiagang Forest Farm, Jiamusi City, Heilongjiang Province, and TLS was used to obtain point cloud data, and trunk analysis was performed after logging. After splicing, cropping, and denoising the point clouds, the iterative random sample consensus (RANSAC) circular fitting algorithm was used to extract the point clouds of individual tree stems. The diameters at different heights on the stems were fitted using the least squares method to generate the stem shape data. A mixed-effect taper equation model was constructed based on the measured stem analysis data. The estimated values of the fitted parameters were used as known variables, while tree height and random effect parameters were treated as the parameters to be estimated. The stem shape data extracted by TLS were fitted tree by tree, and the missing parts of the tree stems were reconstructed using the fitted model. Finally, the tree height, stem shape, and standing timber volume were extracted according to the fitting model and compared with the extraction results obtained without reconstruction. Result: The parameters obtained through tree trunk reconstruction were more accurate than those extracted directly from TLS data. Especially, the estimation error of tree height was significantly reduced, with average deviation and root mean square error percentage reduced by 8.09% and 7.48%, respectively. The proportion of diameter extraction was improved, but there was no significant difference in accuracy. In the estimation of tree volume, the relatively high accuracy of tree volume before and after reconstruction was able to maintained, and the difference was not significant before and after reconstruction with measuremental method by section, and the root-mean-square error percentage of tree volume estimation after reconstruction with binary volume model method was reduced by 4.5%. Conclusion: The method of stem reconstruction by taper equation can effectively restore the missing part of the stem scanned by TLS, which significantly improves the extraction accuracy of TLS individual tree height, and provides new ideas and theoretical support for improving the application accuracy of TLS in forest parameter estimation and forestry survey efficiency.

Key words: terrestrial laser scanning (TLS), taper equation, branch and stem separation, stem reconstruction