数据与计算发展前沿 ›› 2026, Vol. 8 ›› Issue (2): 54-65.

CSTR: 32002.14.jfdc.CN10-1649/TP.2026.02.005

doi: 10.11871/jfdc.issn.2096-742X.2026.02.005

• 专刊：冰冻圈大数据挖掘分析关键技术及应用 • 上一篇    下一篇

基于频域引导特征的泥石流灾后遥感影像语义分割网络构建研究

韩立钦1,2,3(),李龙园2,王钰康2,张耀南3,常盟盟2,*(),潘清元4   

  1. 1 河南师范大学， 地理与旅游学院， 河南 新乡 457003
    2 河南师范大学， 计算机与信息工程学院， 河南 新乡 457003
    3 国家冰川冻土沙漠科学数据中心， 甘肃 兰州 741000
    4 三和数码测绘地理信息技术有限公司， 甘肃 天水 745000
  • 收稿日期:2026-01-14 出版日期:2026-04-20 发布日期:2026-04-23
  • 通讯作者: *常盟盟（E-mail：changmengmeng@htu.edu.cn.）
  • 作者简介:韩立钦，河南师范大学地理与旅游学院，博士，教授，主要研究方向为数据工程防灾减灾、遥感地理信息集成。
    本文主要负责实验方案规划，论文撰写。
    HAN Liqin, Ph.D., is a professor at the School of Geography and Tourism, Henan Normal University. His research interests include data engineering for disaster prevention and mitigation, as well as the integration of remote sensing and geographic information.
    In this paper, he is mainly responsible for experimental design and paper writing.
    E-mail: hanliqin@htu.edu.cn|常盟盟，河南师范大学计算机与信息工程学院，博士，讲师，主要研究方向为时空感知大数据分析及数据智能、城市灾害推断、风险识别与预测等研究。
    本文主要负责算法设计、结果验证评价。
    CHANG Mengmeng, Ph.D., is a lecturer at the School of Computer and Information Engineering, Henan Normal University. His research interests include spatiotemporal big data analytics and data intelligence, urban disaster inference, and risk assessment and prediction.
    In this paper, he is mainly responsible for algorithm design and result verification and evaluation.
    E-mail: changmengmeng@htu.edu.cn
  • 基金资助:
    国家重点研发计划项目(2022YFF0711700);甘肃省科技重大专项(24ZDGE002);天水市科技计划项目(2022-FZJHK-3409)

Research on the Construction of Semantic Segmentation Network for Post-Mudslide Remote Sensing Images Based on Frequency Domain Guided Features

HAN Liqin1,2,3(),LI Longyuan2,WANG Yukang2,ZHANG Yaonan3,CHANG Mengmeng2,*(),PAN Qingyuan4   

  1. 1 School of Geography and Tourism, Henan Normal University, Xinxiang, Henan 457003, China
    2 School of Computer and Information Engineering, Henan Normal University, Xinxiang, Henan 457003, China
    3 National Glaciology, Geocryology and Desert Research Data Center, Lanzhou, Gansu 741000, China
    4 Sanhe Digital Surveying and Mapping Geographic Information Technology, Tianshui, Gansu 745000, China
  • Received:2026-01-14 Online:2026-04-20 Published:2026-04-23

RichHTML

1

PDF

11

摘要：

【目的】 泥石流灾害因其突发性和极大的破坏性成为应急救灾的重要关注点，低空无人机遥感可以高效率获取受灾地区高分辨率影像数据，但如何精准高效地进行受灾区域语义分割一直是研究中面临的共性技术难题。【方法】 本文提出了一种轻量化频率引导遥感分割网络的算法设计方法。【结果】 结果表明：（1）在保持低计算开销的前提下，实现了对高分辨率影像中复杂地物边界结构的高效提取，模型整体性能较U-Net和UKAN方法提升约1.98%和4.43%；（2）快速傅里叶变换（FFT）将空间特征映射至频率域，通过实部与虚部联合建模，有限捕捉远距依赖关系与周期性几何信息，弥补传统卷积在全局建模中的不足；（3）引入全局语义Token，建立了门控融合自适应调控局部特征与全局先验之间的权重，有效缓解了上采样过程中的细节丢失问题。【结论】 研究能够为泥石流灾害快速定损评估提供了有效的技术支撑。

关键词: 语义分割, 深度学习, 频域引导, 泥石流灾害

Abstract:

[Objective] Debris flow disasters, due to their suddenness and extreme destructiveness, have become a key focus in emergency rescue. Low-altitude unmanned aerial vehicle (UAV) remote sensing can efficiently obtain high-resolution image data of disaster-stricken areas. However, how to perform accurate and efficient semantic segmentation of disaster-affected areas remains a common technical challenge in research. [Methods] This paper proposes a lightweight frequency-guided remote sensing segmentation network. [Results] The results show that: (1) Under the premise of maintaining low computational cost, the proposed method efficiently extracts the complex boundary structures of objects in high-resolution images, and the overall performance of the model is improved by approximately 1.98% and 4.43% compared to U-Net and UKAN methods; (2) The Fast Fourier Transform (FFT) maps spatial features to the frequency domain, and by jointly modeling the real and imaginary parts, it effectively captures long-range dependencies and periodic geometric information, compensating for the limitations of traditional convolution in global modeling; (3) The introduction of global semantic tokens establishes a gated fusion mechanism to adaptively regulate the weights between local features and global priors, effectively alleviating the problem of detail loss during the upsampling process. [Conclusions] This research can provide effective technical support for rapid damage assessment of debris flow disasters.

Key words: semantic segmentation, deep learning, frequency-guided features, debris flow disaster