摘要
近年来,随着UAV-RS(无人机遥感)技术迅速发展,该技术已广泛应用在自然资源、林业、海洋渔业等众多领域。该技术是利用UAV-RS和以机器学习为基础的图像识别技术,其具备高精度、高效率和便捷性等诸多优势。现以浙江省舟山衢山岛海域养殖范围鉴定研究项目为例,用SVM(支持向量机)方法计算区域内各功能区域的空间信息,得出区域内的各类特性要素参数。通过监督分类,测试模型模拟效果在不同光谱空间条件下的稳定性,并使用一些样本像素作为训练对象集合。结果显示,SVM在快速地进行地物识别分析的小样本容量条件下,具备了较高的判定性,但在小样本的情况下,其效果受训练集影响较大。因此,各种地物的识别精度可以通过提高样本数据的方法得到提高。
In recent years,the UAV-RS(UAV remote sensing)technology has developed rapidly and been widely applied in natural resources,forestry,and many other fields.The technology is based on UAV-RS and machine learning-based image recognition technology,which has many advantages such as high precision,high efficiency and great convenience.Based on the identification of an aquaculture area in Zhoushan,Qushan Island,Zhejiang Province,SVM(support vector machine)method is used to calculate the spatial information of each functional region,and various characteristic factor parameters in the region are obtained.Through supervised classification,the stability of the model simulation effect under different spectral space conditions is tested,and some sample pixels are used as a set of training objects.The results show that SVM has high determining ability under the condition of small sample size for fast ground object recognition analysis,but meanwhile,its effect is greatly affected by the training set.Therefore,the recognition accuracy of various ground objects can be promoted by improving the sample data.
作者
何复亮
汤湖丁
周玉岑
HE Fuliang;TANG Huding;ZHOU Yucen(Nuclear Geology Brigade of Jiangxi Geological Bureau,Nanchang 330009,China;Shangrao Urban and Rural Planning Research Center,Shangrao 334000,China)
出处
《江西测绘》
2023年第4期17-20,共4页
JIANGXI CEHUI
关键词
UAV-RS
机器学习
支持向量机
监督分类
UAV RS
Machine Learning
Support Vector Machine
Supervised Classification
