基于无人机遥感影像的高精度森林资源调查系统设计与试验

为了实现林业的可持续发展,满足当今森林资源的精准化监测和信息化管理,该文以无人机航拍影像为数据基础,充分结合摄影测量技术、无人机影像后处理技术、地理信息系统技术和林业资源调查管理技术,构建了适用于林业调查和管理的专业森林资源调查系统。该系统以C#为编程语言,结合ArcGIS Engine10.2嵌入式组件技术开发而成,利用无人机影像实现高效快捷的林地空间区划、面积平差和高精度大比例尺的森林小班调查、信息提取等功能,可实现资源数据库的及时更新,极大地缩短了传统调查模式的调查周期,实现了森林资源的科学化管理。以辽宁老秃顶子林场作为试验区,利用1st Opt优化分析软件引入决定系数、估计值的标准差等评定因子,确定试验区冠径、树高、胸径之间的最优模型。同时基于试验区获取数据对系统进行了精度验证,结果表明,该系统获取坡度和高程的相对误差分别为5.17%和5.41%,株树密度、蓄积量的相对误差为2.68%和4.01%。

Design and experiment of high precision forest resource investigation system based on UAV remote sensing images

Abstract: For the sustainable development and the accurate monitoring and information-based management of forestry resources, in this paper, we constructed a forest resource inventory system for forestry investigation and management. This research was based on the unmanned aerial vehicle (UAV) images data with the help of photogrammetric technique, UAV image post-processing technology and geographic information system technology, etc. In the system, we used C# programming language and fully used ArcGIS Engine10.2 embedded component technology. We also used UAV images to achieve efficient functions, such as forest spatial division, area adjustment and forest sub-compartment investigation, vegetation information extraction in high-precision and large-scale. The system can timely update resource database, greatly shorten the traditional investigation period, and achieve the scientific management of forest resources. The extraction of landform factors by software is based on UAV image data, DEM data, and spatial data from spatial division. It can be divided into two steps. The first is to extract the topographical attributes among the UAV image of whole area; the second is to ultimately extract landform factors combined with spatial data to finish attribute tables of each division, which is based on the first step. The goal of the stand factor extraction module is to obtain and store the vegetation information such as sub-compartment volume, stand density, average crown diameter, average tree height, average DBH and canopy coverage. Due to the difficulties for UAV to extract canopy coverage information in the dense forest area, in the study, we divided stand factor extraction module into the sparse forest area and the dense forest area. In the dense forest area, canopy coverage information of each sub-compartment was manually interpreted with the help of pattern spot border, and the input data were directly stored in its corresponding attribute database. In the sparse forest area, based on UAV images and spatial data, the system allowed its users to circle a nearly round sample area within the sub-compartment border, collect information of each single tree in the circle, automatically store the statistical results of all sample information into the corresponding sub-compartment, namely its vegetation information. The empirical study of this forest resource investigation system has been conducted in Lao Tudingzi, Liaoning Province. The results have proven its simple interface, high automation and good interactivity on system operation. The relative error in extraction of slope and elevation were about 5.17% and 5.41% respectively. Before the stand factors were acquired, the relationship between crown diameter and DBH and the relationship between tree height and DBH in test area were fitted by 1stOpt software, then the evaluation indexes, such as coefficient of determination, standard error estimate and total relative error, were introduced to determine the optimal model. The stand factors of the test area have been extracted based on the optimal model with the statistical value of stand density and sub-compartment volume measurement. The results showed the relative error of stand density is 2.68%, and that of sub-compartment volume is 4.01%.