期刊文献+

基于卷积网络的遥感图像建筑物提取技术研究 被引量:15

A Study of Building Extraction from Remote Sensing Imagery Based on Convolution Network
下载PDF
导出
摘要 Mask RCNN是当前最高水平的实例分割算法,本文将该算法应用到高分辨率遥感图像建筑物提取中,提出了一种高效、准确的高分辨率遥感图像建筑物提取方法。首先,利用Tensorflow和Keras深度学习框架搭建Mask RCNN网络模型;然后,通过有监督学习方式在IAILD数据集上进行模型学习。利用训练出的模型对测试集进行建筑物提取实验,通过与基于KNN和SVM等建筑物提取方法对比可以看出,本文方法可以更加完整的、准确的提取出建筑物。采用m AP评价指标对实验结果进行定量评价,本文算法的查全率和查准率均大于对比算法,且多次实验中本文算法的m AP均在81%以上,验证了基于卷积网络的高分辨率遥感图像建筑物提取的有效性和准确性。 Mask RCNN,the state-of-the-art instance segmentation algorithm,is adopted in the building extraction from high-resolution remote sensing imagery.The paper proposes an efficient and accurate building extraction method from high-resolution remote sensing imagery.Firstly,Tensorflow and Keras deep learning framework are used to build the Mask RCNN network model;then,the model learning is carried out on the IAILD data set by supervised learning.By using the trained model,the experiment of building extraction is carried out on the test set.By comparing with the building extraction methods,such as KNN and SVM,it can be seen that this method can be more complete and accurate to extract the building.The m AP evaluation index is used to evaluate the experimental results quantitatively.The recall and precision of the algorithm are all larger than the contrast algorithms,and the m AP of the algorithm is more than 81% in this paper,which verifies the validity and accuracy of the high resolution remote sensing image building extraction based on convolution neural network.
作者 付发 未建英 张丽娜 FU Fa;WEI Jianying;ZHANG Lina(School of Information Engineering,Hebei GEO University,Shijiazhuang 050000,China)
出处 《软件工程》 2018年第6期4-7,共4页 Software Engineering
基金 基于深度学习的遥感图像识别与目标检测(项目编号:CXZZSS2018119)
关键词 深度学习 建筑物提取 MASK RCNN 卷积网络 deep learning building extraction,Mask RCNN CNN
  • 相关文献

二级参考文献8

  • 1曹雪,柯长青.基于对象级的高分辨率遥感影像分类研究[J].遥感信息,2006,28(5):27-30. 被引量:43
  • 2Wang Fangju.Fuzzy supervised classification of remote sensing images[J].IEEE Trans.On Geosci.and Remote Sensing,1990,28(2):194-201.
  • 3Benediktssn J A,Swain P H,Ersoy O K.Neural network approaches versus statistical methods in classification of multisource remote sensing data[J].IEEE Trans.On Geosci.And Remote Sensing,1990,28(4):540-552.
  • 4翟立君.卫星MPLS网络关键技术研究[M].北京:清华大学出版社,2001.
  • 5WINTER Rolf.The Coming of Age of MPLS[J].IEEE Transations on Electron Devices,2011 (59):2964-2972.
  • 6Lucas R,Rowlands A,Brown A,et al.Rule-based Classification of Multi-temporal Satellite Imagery for Habitat and Agricultural Land Cover Mapping[J].ISPRS Journal of Photogrammetry&Remote Sensing,2007(62):165-185.
  • 7邓媛媛,巫兆聪,易俐娜,胡忠文,龚正娟.面向对象的高分辨率影像农用地分类[J].国土资源遥感,2010,22(4):117-121. 被引量:38
  • 8李娜,周德民,赵魁义.高分辨率影像支持的群落尺度沼泽湿地分类制图[J].生态学报,2011,31(22):6717-6726. 被引量:24

共引文献18

同被引文献121

引证文献15

二级引证文献78

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部