| 刘金凤,任晓东,吴 鹏,谭 成,李舒婷,陈 斌,王继雄,饶 昕.基于无人机RGB影像的香格里拉市草原盖度估测研究[J].林业调查规划,2025,50(3):151-158 |
| 基于无人机RGB影像的香格里拉市草原盖度估测研究 |
| Grassland Coverage Estimation in Shangri-La CityBased on UAV RGB Images |
| |
| DOI： |
| 中文关键词: 草原盖度 无人机RGB影像 估测模型 草原类型 香格里拉市 |
| 英文关键词: grassland coverage UAV RGB images estimation model grassland type Shangri-La City |
| 基金项目:云南省林业调查规划院昆明分院科技创新课题;云南省科技厅2023 年第二批科技成果转化专项资金(530000231100001753720). |
|
| 摘要点击次数: 54 |
| 全文下载次数: 48 |
| 中文摘要: |
| 选取香格里拉市82个草原监测样地为研究对象,以无人机RGB影像作为数据源,提取10 种
常用植被指数和HSI颜色特征。采用随机森林回归和Pearson相关性分析,优选CIVE、H和GLI 3个特征,分别采用K最邻近(KNN)、随机森林(RF)、BP神经网络(BPN)和支持向量机(SVM)4 种
算法构建草原盖度估测模型,分析最优模型及草原类型对估测精度的影响。结果表明,RF模型估测精度最高,P>90%,R2>0.8;BPN 和KNN 模型次之,P>80%,R2>0.5;SVM 模型精度最低。RF模
型估测精度总体上与盖度真实值成正相关,且该模型对高寒草甸的估测精度最高,可为相同类型的草原盖度监测提供参考。 |
| 英文摘要: |
| This study selected 82 grassland monitoring sample plots in Shangri-La City as research objects,
using UAV RGB images as the data source to extract 10 common vegetation indices and HSI color
features. Through random forest regression and Pearson correlation analysis, three optimal features—
CIVE, H, and GLI—were selected. Four algorithms, namely K-Nearest Neighbors(KNN), Random
Forest(RF), Backpropagation Neural Network(BPN), and Support Vector Machine(SVM), were employed
to construct grassland coverage estimation models. The study further analyzed the influence of the
optimal model and grassland types on estimation accuracy. The results showed that the RF model achieved
the highest estimation accuracy, with P>90% and R2 >0.8; the BPN and KNN models followed,
with P>80% and R2 >0.5; while the SVM model exhibited the lowest accuracy. The estimation accuracy of the RF model was generally positively correlated with the true coverage values, and this model performed
best for alpine meadows, providing a reference for monitoring grassland coverage of similar types. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
