摘要
随着高空间分辨率遥感影像数据急剧增多,传统的监测分类方法很难适应海量遥感数据的解译需求。本项目以宁夏中卫市沙坡头区光伏电场为例,通过设计适合遥感地物识别的卷积神经网络和构建目标样本的最佳特征表达,让计算机经过深度学习,具备观察和理解的能力,挖掘潜在的有用地物特征,实现光伏电场的自动化信息提取,并且验证了利用高分影像自动化提取的高效性和准确性。
With high spatial resolution remote sensing image data has increased dramatically,the traditional monitoring method is difficult to adapt itself to the requirements of the massive remote sensing data interpretation,the project of central city China area in Ningxia solar electric field as an example,through the design suitable for remote sensing feature recognition of convolution neural network and build the best feature of target sample,let the computer after deep study,observation and understanding ability,excavate the potential useful feature characteristics,realize the automation of photovoltaic electric field information extraction,and validate the use of high efficiency and accuracy of image automatic extraction.
作者
严瑾
梁伟
闫亭廷
刘长宁
YAN Jin;LIANG Wei;YAN Tingting;LIU Changning(Ningxia Hui Autonomous Region Remote Sensing Surveying and Mapping Exploration Institute,Ningxia Hui Autonomous Region Remote Sensing Center,Yinchuan 750021,China)
出处
《测绘与空间地理信息》
2021年第8期75-77,共3页
Geomatics & Spatial Information Technology
基金
宁夏回族自治区重点研发计划一般项目——宁夏遥感大数据平台研发及生态环境监测应用研究(2019BDE03006)。
关键词
高分影像
自动化
光伏
模型
high resolution image
automation
photovoltaic(pv)
model