| 地面成像光谱数据的田间杂草识别 |
| |
| 引用本文： | 李颖,张立福,严薇,黄长平,童庆禧.地面成像光谱数据的田间杂草识别[J].遥感学报,2013,17(4):855-871. |
| |
| 作者姓名： | 李颖 张立福 严薇 黄长平 童庆禧 |
| |
| 作者单位： | 中国气象局河南省农业气象保障与应用技术重点开放实验室/河南省气象科学研究所,中国科学院遥感应用研究所,61363部队,中国科学院遥感应用研究所,中国科学院遥感应用研究所 |
| |
| 基金项目： | 国家公益性行业(气象)科研专项(编号:GYHY200906022);中国气象局河南农业气象保障与应用技术重点开放实验室开放研究基金(编号:AMF201207) |
| |
| 摘 要： | 地面成像光谱数据兼具高光谱分辨率与高空间分辨率,在田间杂草识别中具有很好的应用前景。目前基于机器视觉的杂草识别方法以形状特征为主,当作物杂草形态相似时识别的困难和利用高光谱特征以像元为单元识别时效率较低,不利于实时自动化除草,因此,本文提出一种综合面向对象与高光谱特征匹配的杂草识别方法,在对作物杂草对象样本的形状特征和光谱曲线提取分析的基础上,建立基于形状特征规则与光谱角匹配的植物对象识别决策树,用于识别实验田中的作物杂草对象。实验结果表明,当场景中某些不同种类植物对象的形态相似时,基于形状特征规则与光谱角匹配的杂草识别方法可借助高光谱特征精细区分植物对象的种类,且在形状特征规则约束下使用高光谱特征匹配法识别植物对象,可克服"同物异谱"和"同谱异物"现象带来的不确定性,该方法识别精度可优于仅使用光谱角匹配法的情况,并优于使用颜色和形状分析技术的情况。
|
| 关 键 词： | 地面成像光谱数据 杂草识别 面向对象 形状特征 光谱角匹配 |
| 收稿时间： | 2012/7/31 0:00:00 |
| 修稿时间： | 2012/11/16 0:00:00 |
Weed identification using imaging spectrometer data |
| |
| LI Ying,ZHANG Lifu,YAN Wei,HUANG Changping and TONG Qingxi.Weed identification using imaging spectrometer data[J].Journal of Remote Sensing,2013,17(4):855-871. |
| |
| Authors: | LI Ying ZHANG Lifu YAN Wei HUANG Changping and TONG Qingxi |
| |
| Institution: | CMA Henan Key Laboratory of Agrometeorological Support and Applied Technique/Henan Institute of Meteorological Sciences,Institute of Remote Sensing Applications, Chinese Academy of Science,61363 Troops,Institute of Remote Sensing Applications, Chinese Academy of Science,Institute of Remote Sensing Applications, Chinese Academy of Science |
| |
| Abstract: | Weed identification is a basic task of precision agriculture, as the premise of variable spraying and accurately herbicidal. Field imaging spectrometer data possessing both high spectral resolution and high spatial resolution has an application prospect in weed identification. Currently, for weed identification, the methods mainly considering shape features based on machine vision perform poorly when weeds and crops have the similar shape features, while the methods using high spectral characteristics have low efficiency when it is identified in pixels. In order to overcome the shortcoming of existing methods, a new weed identification approach combining the strengths of both object oriented idea and spectral characteristic matching idea was proposed. The proposed approach extracts and analyzes the shape features and the spectral curves of plant object samples, and then builds a decision tree using shape feature rules and spectral angle mapper to identify the plant objects in the experimental field. The result showed that the proposed approach could identify different kinds of plant objects which have similar shape by using hyperspectral characteristics, and could overcome the identification difficulties when same object with different spectra or different objects with same spectrum by using shape feature rules. The identification accuracy of the described approach is higher than spectral angle mapper method and the color and shape analysis techniques. |
| |
| Keywords: | weed identification object oriented shape feature spectral angle mapper |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《遥感学报》浏览原始摘要信息 |
| 点击此处可从《遥感学报》下载免费的PDF全文 |
|
