【作者】 黄万里；

【导师】 李虎；

【作者基本信息】 福建师范大学 ， 地图学与地理信息系统， 2015， 博士

【摘要】 森林小班区划是森林资源调查中的首要任务和主要内容,也是森林资源管理与动态监测的主体业务,在林业生产经营及林业与森林生态环境等科学研究中有着极其重要的意义。从目前的相关研究看,由于森林固有的多尺度特点,及小班区划时需要考虑来自不同尺度的森林地物信息,加之,高空间分辨率遥感图像上的光谱信息受限,小班区划的技术指标在图像上的指示信息有限,采用单一尺度的图像分割与分类信息进行小班边界提取难以取得满意的效果。基于此,本文在构建多尺度层次图像分割方法并进行最优分割尺度选择的基础上,构建了基于对象基元与林分尺度分类结果的林分分类指标,并进行林分分类与小班边界提取,力图探索一种高分辨率遥感图像小班边界提取的新思路。选择新疆天山西部林业局的特克斯林场为研究区,按照“森林信息的图像表征→图像多尺度分割与信息提取→森林小班对象识别”开展面向山地森林的高空间分辨率遥感小班边界提取研究。针对天山林区展开实验,结果表明,该方法对于实验区的林分分类及小班边界的提取可以取得比较好的结果。本文提出的林分分类的多尺度结构指数不但可以在林业上进行林分分类、小班边界提取,还可能用于城市建筑区或居住区的划分、分类等。更多还原

【Abstract】 Sub-compartment division is the one of the primary tasks and major contents in the forest resource inventory, as well as the main business of the forest resources management and dynamic monitoring, which plays a vital role in the scientific researches, such as the forestry production and management and forest eco-environment. A review of the current studies, the specifications of the sub-compartment division is quite limited in the indicator information of image, and it is difficult to achieve satisfactory result of sub-compartment boundary with single-scale image segmentation and classified information, due to the disadvantages, such as the inherent multi-scale features of the forest, the consideration of different scales of forest features in the sub-compartment division, limited spectral information of the high spatial resolution remote sensing image (HSRRSI). In view of these limitations, the stands classification index is constructed based on the object primitive and stands level classification result, and the stands classification and sub-compartment boundary extraction (SBE) are carried out for trying to explore a new thought of the SBE of the HSRRSI, on the basis of the construction of multi-scale segmentation method and optimal segmentation scale selection.The Tekesi forest farm of Xinjiang province, P. R. China is taken as the study area, and the study on the HSRRSI SBE of the forest is launched by following "image representation of forest information→ multi-scale image segmentation and information extraction→ sub-compartment recognition". Experiments are conducted in Tianshan forest, and according to the result, the method is favorable for the stands classification and sub-compartment extraction of the experiment region. The multi-scale structural index of the stands classification proposed in this dissertation can not only carry out the stands classification, SBE in the forest industry, but also be applied to the division, classification of the urban building zone or residence zone, etc.更多还原