基于CNN的InSAR干涉图质量评价方法

A Quality Evaluation Method for InSAR Interferograms Based on Convolutional Neural Network

ES评分 0

DOI 10.14075/j.jgg.2025.04.132
刊名
Journal of Geodesy and Geodynamics
年,卷(期) 2026, 46(3)
作者
作者单位

1 江西师范大学地理与环境学院, 南昌, 330022;
2 江西省地震局, 南昌, 330026

摘要
基于卷积神经网络(CNN)提出一种InSAR干涉图质量的自动评价方法, 并将该方法应用于Sentinel-1数据实验中, 测试集的准确率和召回率分别达到96.5%和98%, 自动筛选与人工挑选结果具有很好的一致性。将该方法应用于南昌地区PS-InSAR形变处理, 结果表明, 通过质量评价选择高质量的干涉图开展形变分析, 能够显著提高形变时间序列测量精度。
Abstract
Based on convolutional neural network (CNN), an automatic evaluation method for InSAR interferogram quality is proposed. This method is applied to Sentinel-1 data experiments, achieving accuracy and recall rate of 96.5% and 98% on the test set, respectively. The results of automatic screening and manual selection have good consistency. We apply the method to PS-InSAR deformation processing in Nanchang area, it is found that selecting high-quality interferograms for deformation analysis through quality evaluation can significantly improve the accuracy of deformation time series measurement.
关键词
CNN; 时序InSAR; 干涉图质量; StaMPS; PS-InSAR
KeyWord
CNN; time series InSAR; interferogram quality; StaMPS; PS-InSAR
基金项目
页码 360-367
  • 参考文献
  • 相关文献
  • 引用本文

查小惠, 陈皆红, 高华, 齐述华. 基于CNN的InSAR干涉图质量评价方法 [J]. 大地测量与地球动力学. 2026; 46; (3). 360 - 367.

  • 文献评论

相关学者

相关机构