地表植被作为人类自然生态系统的一个重要组成部分,在减轻土壤侵蚀、维持生态平衡、提高防风固沙能力,以及维持局部区域经济可持续发展等许多方面都发挥着重要的主导作用.近年来,卫星遥感技术凭借其大范围、多尺度、多时相的优势,已经成为一种低成本、高效率的植被覆盖估计算法,在植被提取和变化监测中表现出了重要的应用价值.论文以榆林市毛乌素沙漠地区为主要研究区域,利用手机可视化虚拟图像处理环境(Environment for Visualizing Images,ENVI)分析软件对landsat8遥感影像进行地表植被的信息提取,采用不同的监督分类方法对植被信息进行分类提取,并对不同的监督分类结果进行质量精度分析评定.结果表明,神经网络方法较其他方法在沙漠植被提取方面表现出较优的性能,可以实现对沙漠植被的动态变化监测.
Research on Extraction Method of Desert Vegetation Based on Neural Network
As an important part of the human natural ecosystem,surface vegetation plays an important leading role in many as-pects,such as reducing soil erosion,maintaining ecological balance,improving windbreak and sand fixation capacity,and main-taining sustainable regional economic development.In recent years,satellite remote sensing technology has become a low-cost and high-efficiency vegetation cover estimation algorithm with its advantages of large-scale,multi-scale and multi-temporal phases,and has shown important application value in vegetation extraction and change monitoring.In this paper,the Mu Us Desert area of Yulin city is taken as the main research area,and the information of surface vegetation is extracted from landsat8 remote sensing im-ages by using the environment for visualizing images(ENVI)analysis software of mobile phones,and the vegetation information is classified and extracted by different supervised classification methods,and the quality accuracy of different supervised classification results is analyzed and evaluated.The results show that the neural network method has better performance than other methods in des-ert vegetation extraction,and can realize the dynamic change monitoring of desert vegetation.
国家科技期刊平台
NSTL
万方数据
