论文部分内容阅读
平原绿化是我国平原地区森林植被的主体,为生态环境建设提供了有力保障。但平原绿化分布零散,斑块面积小,不易被准确提取。本文以河南省封丘县为例,利用天绘一号卫星影像面向对象分类技术提取了平原绿化。在面向对象分类过程中,针对不同平原绿化信息选取不同的分割尺度和参数,建立影像对象的分割层次结构,确保农田林网、通道防护林、四旁防护林等平原绿化信息能够被准确分割。分类时通过构建隶属度函数,利用层与层之间的关系,子类继承父类的分类特征,在此基础上增加自己的特征,建立分类规则集,逐层提取绿化信息。本文共获取封丘县平原绿化面积152.51 km2,其中,农田林网36.09 km2,通道防护林21.29 km2,四旁防护林71.56 km2,片林23.57 km2。利用验证点数据,对分类结果进行精度评价,总体分类精度为93.50%,Kappa系数达到0.92。
Plain greening is the main body of forest vegetation in the plain areas of our country, which provides a powerful guarantee for the ecological environment construction. However, scattered plains green, plaque area is small, not easy to be accurately extracted. In this paper, Fengqiu County of Henan Province as an example, the use of day-painted satellite image-based object-oriented classification technology extracted plain green. In the process of object-oriented classification, different segmentation scales and parameters are selected according to different greening information of the plain, and the hierarchical structure of the image objects is established to ensure that the greening information of plain, shelterbelts, shelterbelts and other four shelters can be accurately partitioned. By constructing the membership function and using the relationship between layers, the subclasses inherit the classification features of the parent class. Based on this, subclasses add their own features to establish the classification rules set and extract the green information layer by layer. In this paper, a total of 152.51 km2 of green area in Fengqiu County was obtained, of which 36.09 km2 was farmland forest net, 21.29 km2 forest shelterbelts, 71.56 km2 shelterbelts and 23.57 km2 forest. Using the verification point data, the accuracy of the classification results was evaluated. The overall classification accuracy was 93.50% and the Kappa coefficient was 0.92.