王利军,郭 燕,贺 佳,王利民,张喜旺,刘 婷.基于决策树和SVM的Sentinel-2A影像作物提取方法[J].农业机械学报,2018,49(9):146-153.
WANG Lijun,GUO Yan,HE Jia,WANG Limin,ZHANG Xiwang,LIU Ting.Classification Method by Fusion of Decision Tree and SVM Based on Sentinel-2A Image[J].Transactions of the Chinese Society for Agricultural Machinery,2018,49(9):146-153.
摘要点击次数: 2301
全文下载次数: 1158
基于决策树和SVM的Sentinel-2A影像作物提取方法   [下载全文]
Classification Method by Fusion of Decision Tree and SVM Based on Sentinel-2A Image   [Download Pdf][in English]
投稿时间:2018-03-29  
DOI:10.6041/j.issn.1000-1298.2018.09.017
中文关键词:  秋季作物  Sentinel-2A  植被指数  决策树  支持向量机
基金项目:国际科技合作项目(182102410024)、国家重点研发计划项目(2016YFD0300609)、国家自然科学基金项目(41601213)和河南省重大科技专项(171100110600)
作者单位
王利军 河南省农业科学院 
郭 燕 河南省农业科学院 
贺 佳 河南省农业科学院 
王利民 中国农业科学院 
张喜旺 河南大学 
刘 婷 河南省农业科学院 
中文摘要:以河南省濮阳县为研究区,以2017年8月6日遥感影像为基础数据源,基于地面样方和样本点数据分析构建植被指数阈值分割分类决策树,结合支持向量机(Support vector machine,SVM)分类方法实现了秋季主要作物种植面积遥感识别,并与其他方法分类结果进行了精度验证与对比。结果表明,与最大似然法(Maximum likelihood,ML)和SVM法相比较,决策树和SVM相结合能较好地解决线状地物和小地块作物提取不全以及“椒盐”现象等问题,可以对秋季复杂作物进行有效识别,作物分类提取总体精度和Kappa系数分别为92.3%和0.886。利用中分辨率单时相遥感影像,结合波谱特征和植被指数能有效提高复杂作物分类精度,为区域复杂作物分类提取提供技术参考和借鉴价值。
WANG Lijun  GUO Yan  HE Jia  WANG Limin  ZHANG Xiwang  LIU Ting
Henan Academy of Agricultural Sciences,Henan Academy of Agricultural Sciences,Henan Academy of Agricultural Sciences,Chinese Academy of Agricultural Sciences,Henan University and Henan Academy of Agricultural Sciences
Key Words:autumn crops  Sentinel-2A  vegetation index  decision tree  support vector machine
Abstract:Remote sensing images with medium spatial resolution and multiband can provide data source for crop classification on country scale. Based on analysis of the spectral characteristics of single image and its vegetation index, the identification and acreage extraction of major autumn crops can be effectively achieved. Taking Puyang County, Henan Province as study area, and basic image with 13 bands and spatial resolution of 10m, which was collected on August 6th, 2017 was employed. Combined with the ground samples and sample points data, the spectrum curve characteristics, including NDVI and RENDVI of the major crop types (corn, peanut, soybean, rice and minor crops such as vegetables, sweet potato, etc.) in this growth period were extracted. Through the analysis of spectrum curve characteristics of different crop types, it can divide the basic image into different regions by building decision tree, which was built by the threshold segmentation of vegetation index features, and band math tool in ENVI software. Based on curve characteristics of NDVI, the basic image data can tripartite regions such as major crop planting region, non crop planting region and minor crops planting region. Then on the basis of major crop planting region image and its RENDVI data, it can divide this region into two regional images, including corn/rice and peanut/soybean. Finally, synthesizing the above results, five crop types in the study area were classified by SVM and limited training samples. The precision of the results by using decision tree and SVM was evaluated compared with ML and SVM methods, which were gradually adjusted according to the validation of field samples and sample points. The method can effectively solve problems such as incomplete extraction of linear object, different crops in small plots, and also phenomena of “salt and pepper”. Its overall accuracy and Kappa coefficient reached 92.3% and 0.886, respectively. The precision of classification can meet the demand of remote sensing image classification by analyzing spectral characteristics and vegetation index. Based on mono temporal Sentinel-2A data, it can provide data support and technical reference for regional complex crop classification extraction.

Transactions of the Chinese Society for Agriculture Machinery (CSAM), in charged of China Association for Science and Technology (CAST), sponsored by CSAM and Chinese Academy of Agricultural Mechanization Science(CAAMS), started publication in 1957. It is the earliest interdisciplinary journal in Chinese which combines agricultural and engineering. It always closely grasps the development direction of agriculture engineering disciplines and the published papers represent the highest academic level of agriculture engineering in China. Currently, nearly 8,000 papers have been already published. There are around 3,000 papers contributed to the journal each year, but only around 600 of them will be accepted. Transactions of CSAM focuses on a wide range of agricultural machinery, irrigation, electronics, robotics, agro-products engineering, biological energy, agricultural structures and environment and more. Subjects in Transactions of the CSAM have been embodied by many internationally well-known index systems, such as: EI Compendex, CA, CSA, etc.

   下载PDF阅读器