摘要
随着经济快速发展,工业固废、非正规垃圾、未覆盖建筑渣土等固体废物急剧增加,对区域生态环境造成极大威胁。固废堆场具有面积小、分布散等特点,目前国内仍缺少针对各类固废堆场的遥感自动识别研究。为此,基于国产高分辨率卫星遥感数据,根据野外实地光谱采集结果,分别开展了未覆盖建筑渣土、工业固废及非正规垃圾的自动识别方法研究,提取研究区各类固废堆场。结果表明:未覆盖建筑渣土在蓝波段与绿波段分别存在“吸收谷”与“反射峰”,基于该特征构建的比值指数模型,结合直方图双峰法阈值分割可以有效提取未覆盖建筑渣土区域;工业固废、非正规垃圾2种固废类型多样,光谱反射率没有明显规律,结合其纹理、色调等特征,采用面向对象多尺度分割、支持向量机监督分类方法能够较好地识别2种类型固废;基于自动化提取技术并结合人机交互判读方法,提取的研究区未覆盖建筑渣土、工业固废及非正规垃圾等3种固废堆场的精度分别达到96.83%、88.26%、85.71%,各类固体废物遥感识别精度较高,极大提高了固废监测效率。
With the rapid economic development,solid wastes such as industrial solid waste,informal waste and uncovered construction residue increase rapidly,which poses a great threat to the regional ecological environment.Solid waste yards are characterized by small area and scattered distribution.At present,there is a lack of research on automatic remote sensing identification of all kinds of solid waste yards in China.Therefore,based on domestic high-resolution satellite remote sensing data and field spectral acquisition results,automatic identification methods of uncovered construction residue,industrial solid waste and informal waste were studied and various solid waste dumps in the study area were extracted.The results showed that:There were“absorption valleys”and“reflection peaks”of uncovered construction residue in blue band and green band respectively.The ratio index model constructed based on this feature combined with histogram bimodal threshold segmentation could effectively extract the area of uncovered construction residue.There were various types of industrial solid waste and informal solid waste,and the spectral reflectance was not obvious.Object-oriented multi-scale segmentation and support vector machine supervised classification method could better identify the two types of solid waste from the perspectives of texture and tone.Based on automatic extraction technology of solid waste and human-computer interaction interpretation method,the accuracy of the three kinds of solid waste storage sites in the study area,including uncovered construction residue,industrial solid waste and informal waste,reached 96.83%,88.26%and 85.71%respectively.Remote sensing identification accuracy of all kinds of solid waste was high,which greatly improved the efficiency of solid waste monitoring.
作者
陈绪慧
李静
申文明
史雪威
蔡明勇
薛志刚
闫文生
CHEN Xuhui;LI Jing;SHEN Wenming;SHI Xuewei;CAI Mingyong;XUE Zhigang;YAN Wensheng(Satellite Application Center for Ecology and Environment,Ministry of Ecology and Environment,Beijing 100094,China;Department of Soil and Solid Waste Chemicals,Chengdu Ecological Environment Bureau,Chengdu 610042,China;Chengdu Ecological Environment Digital Intelligence Management Center,Chengdu 610015,China)
出处
《中国环境监测》
CAS
CSCD
北大核心
2023年第6期218-226,共9页
Environmental Monitoring in China
基金
高分辨率对地观测重大专项(05-Y30B01-9001-19/20-2)。
关键词
固废
高分辨率影像
比值指数
面向对象多尺度分割
支持向量机监督分类
solid waste
high-resolution images
ratio index
object-oriented multiscale segmentation
support vector machine supervised classification
