基于无人机影像天山云杉林主伐迹地提取研究

【目的】通过分析天山云杉主伐迹地更新群落特征,将其划分为三个等级的主伐迹地,基于无人机影像,以实现提取不同等级主伐迹地的面积以及确定其分布,为新疆天保工程实施后山区森林资源更新恢复及评价提供科学依据。【方法】采用eCognition软件面向对象多尺度分割技术对研究区进行分割,通过最邻近分类以及SEaTH算法相结合的分类方法筛选分类特征,并建立类别层次结构及分类规则集,以实现提取不同等级主伐迹地面积,并进行精度评价。【结果】利用最邻近分类方法与SEaTH算法相结合的分类方法提取不同等级的主伐迹地,其总体分类精度达到81.82%,Kappa系数为0.74。提取主伐迹地Ⅰ、Ⅱ、Ⅲ的面积吻合度分别为87.09%、79.86%和66.33%;提取主伐迹地总面积为72.574 3 hm²,目视解译面积为92.174 9 hm²,面积相对误差为21.26%,面积吻合度为78.74%。【结论】该方法用于研究区提取主伐迹地信息是可行的。主伐迹地Ⅰ提取的面积吻合度最高,由于主伐迹地Ⅰ内的天山云杉林分株数密度较低,其林分平均冠幅、平均年龄等因子均相对较小,因此在影像上更好区分。主伐迹地Ⅱ分布面积最大,主伐迹地Ⅰ和Ⅲ分布面积较小,研究区内主伐迹地林分生长状况主要为主伐迹地Ⅱ状态

Research on Extraction of Final Felling Area of Picea Schr enkiana var tianshanica Based on UAV Image

【Objective】 Based on the analysis of the characteristics of the regeneration community of the final felling area of Picea Schrenkiana var tianshanica in Tianshan Mountain, it was divided into three grades of final felling area. Based on UAV images, the area and distribution of the final felling area of different grades can be extracted and determined in the hope of providing technical support for the evaluation of forest resources renewal and restoration in mountainous areas after the implementation of Xinjiang natural forest conservation project to a certain extent. 【Method】The object-oriented multi-scale segmentation technology of eCognition software was used to segment the study area. Through the combination of nearest neighbor classification and SEaTH algorithm, the classification features were screened, and the classification hierarchy and classification rule set were established so as to realize the extraction of the area of the main cutting area of different grades, and carry out the precision evaluation.【Result】The nearest neighbor classification method combined with the SEaTH algorithm was used to extract the main logging sites of different grades. The overall classification accuracy was 81.82% and the Kappa coefficient was 0.74. The area coincidence of the final felling areaⅠ, Ⅱ and Ⅲ were 87.09%, 79.86% and 66.33%. The total area of the final felling area was 72.574,3 hm², visual interpretation area was 92.174,9 hm², area relative error was 21.26% and area coincidence was 78.74%.【Conclusion】It is feasible to use this method to extract the information of final felling area in the study area. Because of the lower number density of Picea Schrenkiana var tianshanica, the average crown width and average age of the spruce stand are relatively small, so it is better to distinguish in image. The distribution area of the main cutting site II is the largest, while the distribution area of the main cutting site Ⅰ and III is small. The stand growth status of the main cutting land in the study area is mainly in the second state of the cutting site.