摘要
为探索GF-1 PMS多光谱数据影像在低山丘陵地貌破碎地区主要农作物遥感识别中的信息有效性,以重庆市永川区卫星湖街道为例,利用研究区多时相、多光谱特征影像,对研究区油菜、玉米、水稻等主要作物进行信息提取。提取结果显示,利用GF-1多时相多特影像,水稻作物信息提取生产精度与用户精度均达到90%以上,提取精度较低的旱地作物玉米提取效果也得到了提升,油菜作物信息提取生产精度大幅度高于用户精度,主要作物提取总体精度OA为80.93%,Kappa系数0.635,分类质量达到较好水平。基于多时相GF-1影像光谱、纹理等特征的面向对象分类方法,能够有效地提取南方低山丘陵破碎地貌地区主要农作物空间分布信息,提高主要农作物的遥感识别精度,为山地农作物遥感信息提取提供参考。
In order to explore the information validity of GF-1 PMS multi-spectral data images in remote sensing identification of main crops in low hilly and broken areas,taking the Satellite Lake Street in Yongchuan District as an example,using multi-temporal and multi-spectral features,The main crops such as rape,corn and rice were extracted.The extraction results show that the yield precision and user precision of rice were effectively improved,reaching more than 90%,and the extraction efficiency of dryland crops with low extraction precision was also improved;the production precision of rape was significantly higher than that of user accuracy,the overall crop extraction accuracy is 80.93%,Kappa coefficient is 0.635,and the classification quality reaches a good level.The object-orien-ted classification method based on multi-temporal GF-1 image spectrum,texture and other features can effectively extract the spatial distribution information of main crops in the broken mountainous terrain of low mountains in the south,improve the remote sensing rec-ognition accuracy of main crops,and provide remote sensing information extraction for mountain crops reference.
作者
王克晓
周蕊
王茜
虞豹
黄祥
WANG Kexiao;ZHOU Rui;WANG Qian;YU Bao;HUANG Xiang(Chongqing Academy of Agricultural Sciences,Chongqing 401329,China)
出处
《测绘与空间地理信息》
2020年第6期33-36,共4页
Geomatics & Spatial Information Technology
基金
重庆市技术创新与应用发展专项(cstc2019jscx-msxmX0355)资助。