摘要
广西甘蔗种植区域离散,因混杂于多种农作物中,其光谱易受其他作物的影响,故利用单一时相多光谱遥感影像提取甘蔗有一定的困难。针对这一难题,该文首先提出甘蔗最佳识别时段,基于多时相HJ-1A/1B星CCD影像,以广西中部贵港市三区为研究区,通过面向对象分类软件eCognition,利用甘蔗在不同时相影像上的光谱特征:光谱均值、归一化植被指数NDVI和由灰度共生矩阵导出的局部一致性指数GLCM homogeneity,建立决策树逻辑的分类规则集提取甘蔗种植区。结果表明该方法能较精确地进行甘蔗识别,最大程度消除其他干扰因素影响,分类精度为91.3%,kappa系数为0.83,同时也证实了HJ卫星CCD多光谱遥感数据应用于甘蔗识别的有效性。
Sugarcane identification on specific parcels and the assessment of soil management practices are important for agro-ecological studies, greenhouse gas modeling, and agrarian policy development. Information on the sugarcane cultivation areas is of global economic and environmental significance. The study area is Guigang City located in the central area of Guangxi Province which is a good representation of the agricultural conditions. Traditional pixel-based analysis of remotely sensed data results in inaccurate identification of some crops due to pixel heterogeneity, mixed pixels, spectral similarity. The growing region of sugarcane in Guangxi Province is discrete, so the remote sensing spectral of sugarcane is vulnerable to be impacted on a variety of crops. There are certain difficulties in the use of multi-spectral remote sensing to extract sugarcane. Current techniques for mapping sugarcane are based mainly on MODIS satellite data and may not make full use of the texture characteristics. The objective of this research is to investigate the potential for the application of the China Environment Satellite HJ-1A/1B and Phenology in monitoring sugarcane cultivation areas in Guangxi province in southern China. In our approach, we explored several characteristics such as the time information, spectral characteristics and texture features, used an object-based image analysis method and decision tree method for mapping the sugarcane area over large areas based on multi-temporal China Environment Satellite HJ-1A/B Data. A CCD camera sensor with 30m spatial resolution on board the China Environment Satellite HJ-1A and B has both visible and near infrared bands and a revisit period of four days, thus the temporal Normalized Difference Vegetation Index（NDVI） can be obtained from HJ-1A and B data. A time series of the China Environment Satellite HJ-1A/B Data and DEM images was acquired in order to represent the wide range of pattern variation along the sugarcane crop cycle. Firstly, the phenology differences and betwee
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2014年第11期145-151,共7页
Transactions of the Chinese Society of Agricultural Engineering
基金
中国科学院战略性先导科技专项--应对气候变化的碳收支认证及相关问题(XDA05050107)
关键词
遥感
植被
分类
甘蔗
面向对象
物候特征
remote sensing
vegetation
classification
sugarcane
object-oriented
phenology
