基于无人机影像的天山云杉林单木树冠信息提取

The Extraction of Single Wood Canopy of Picea Schrenkiana var tianshanica Forest Based on UAV Image

【目的】研究树冠信息估测出林分密度、生长量等森林调查指标,判断林木生长优良状况提供参考,【方法】基于无人机遥感影像,以新疆农业大学实习林场主伐迹地下天山云杉林(Picea Schrenkiana var tianshanica)为研究对象,采用高斯-拉普拉斯算子(Laplacian of Gaussian,LOG)结合最大类间方差寻找最优阈值(Otsu)对影像进行处理,并利用标记控制分水岭分割方法分别提取疏、中、密3种不同郁闭度的天山云杉单木树冠信息。【结果】利用优化后的标记控制分水岭分割方法较好的解决了过分割问题,对单木树冠信息提取的F测度在疏、中、密林区分别是98.26%、92.91%和87.57%。【结论】使用的方法提取单木树冠信息精度较高,可以评价对天山云杉林的生长状况,可对主伐迹地下天山云杉林的更新和恢复提供可靠的技术支撑。

【Objective】The tree crown growth status can be used as an important index to judge the good growth of trees,and the extraction of tree crown information can estimate forest survey indexes such as stand density and growth quantity.【Methods】Based on the remote sensing image of UAV,the underground Picea Schrenkiana var tianshanica forest in Xinjiang Agricultural University Internship Forest Farm was taken as the research object.LOG combined with the maximum inter-class variance was used to find the optimal threshold(Otsu)to process the image,and the mark-controlled watershed segmentation method was applied to extract the crown information of the single tree of Picea Schrenkiana var tianshanica with three different depression densities.【Results】The results showed that the optimized mark-controlled watershed segmentation method could solve the problem of over-segmentation,and the F measurements extracted from the crown information of the single tree were 98.26%,92.91%and 87.57%,respectively in the sparse,medium and dense forest areas.【Conclusion】The method used in this paper has high accuracy in extracting single tree crown information,which can better evaluate the growth status of Picea Schrenkiana var tianshanica forest,and provide reliable technical support for the regeneration and restoration of Picea Schrenkiana var tianshanica forest.