基于无人机遥感影像的水稻种植信息提取研究
Research on Rice Planting Information Extraction Based on UAV Remote Sensing Images
林弘晖 1林腾2
作者信息
- 1. 武汉大学测绘学院,武汉,430079
- 2. 福建省地质测绘院,福州,350011
- 折叠
摘要
基于无人机遥感影像,通过差异系数优选出过绿指数(EXG)、可见光波段差异植被指数(VDVI)、归一化绿蓝差异指数(NGBDI)等3个植被指数特征,与B波段相关性、G波段协同性的2个纹理特征,结合实地调查数据,采用支持向量机回归分类算法进行地物分类;提取小湖镇小湖村的水稻种植面积,并进行精度评价.研究表明:结合光谱特征、植被指数和纹理特征进行地物分类效果最佳,其总体分类准确度为84.19％,Kappa系数为0.801 8％,水稻的制图精度和用户精度分别为80.77％和88.42％.该方法为水稻种植信息提取提供了一种准确有效的技术途径和应用价值.
Abstract
Based on UAV remote sensing imagery,three vegetation index features were selected through the difference coefficient:Excess Green Index(EXG),Visible Difference Vegetation Index(VDVI),and Normalized Green-Blue Difference Index(NGBDI),and two texture features of the correlation of Band B and the coherency of Band G.By combining these features with field survey data and employing a Support Vector Regression(SVR)classification algorithm,a classification of ground objects was conducted to extract rice information in Xiahu Town,Xiahu Village.The accuracy of the classification was then evaluated.The study shows that the best results for land cover classification are achieved by combining spectral features,vegetation indices,and texture features,with an overall classification accuracy of 84.19％and a Kappa coefficient of 0.8018.The mapping accuracy and user's accuracy for rice are 80.77％and 88.42％,respectively.This method provides an accurate and effective technical approach andapplication value for extracting rice planting information.
关键词
无人机影像/特征选择/分类算法/水稻种植信息Key words
UAVRS/feature selection/classification/rice planting information引用本文复制引用
出版年
2024
NETL
NSTL
万方数据