面向洪水灾害的视觉-文本协同表征的异质遥感变化检测方法

doi:10.11947/j.AGCS.2026.20250521

测绘学报 ›› 2026, Vol. 55 ›› Issue (5): 927-940.doi: 10.11947/j.AGCS.2026.20250521

面向洪水灾害的视觉-文本协同表征的异质遥感变化检测方法

尉锐¹(), 李杰¹^,², 刘汇慧¹(), 吴美茹¹, 林镠鹏³, 袁强强¹, 郑莉¹

^1.武汉大学测绘学院，湖北　武汉　430079
^2.湖北珞珈实验室，湖北　武汉　430079
^3.武汉大学资源与环境科学学院，湖北　武汉　430079

收稿日期:2025-12-15 修回日期:2026-04-21 出版日期:2026-06-23 发布日期:2026-06-23
通讯作者: 刘汇慧 E-mail:rui.yu@whu.edu.cn;hhliu@sgg.whu.edu.cn
作者简介:尉锐（2002—），男，硕士生，研究方向为遥感变化检测、深度学习多任务联合。E-mail：rui.yu@whu.edu.cn
基金资助:
国家自然科学基金(42471504; 42301417)

Heterogeneous remote sensing change detection based on vision-language collaborative representation for flood disasters

Rui YU¹(), Jie LI¹^,², Huihui LIU¹(), Meiru WU¹, Liupeng LIN³, Qiangqiang YUAN¹, Li ZHENG¹

^1.School of Geodesy and Geomatics, Wuhan University, Wuhan 430079, China
^2.Hubei Luojia Laboratory, Wuhan 430079, China
^3.School of Resource and Environmental Sciences, Wuhan University, Wuhan 430079, China

Received:2025-12-15 Revised:2026-04-21 Online:2026-06-23 Published:2026-06-23
Contact: Huihui LIU E-mail:rui.yu@whu.edu.cn;hhliu@sgg.whu.edu.cn
About author:YU Rui (2002—), male, postgraduate, majors in remote sensing change detection and deep learning-based multi-task learning.　E-mail: rui.yu@whu.edu.cn
Supported by:
The National Natural Science Foundation of China(42471504; 42301417)

摘要/Abstract

摘要：

光学与SAR影像的异质变化检测在灾害应急响应与全天候监测中具有重要意义。然而，二者成像机理的显著差异导致特征分布不一致，加之缺乏标注样本与文本描述，制约了传统方法及现有深度学习模型的检测性能。为此，本文提出了一种多维度变化特征增强的CLIP变化检测网络（MCE-CLIP），聚焦于解决洪水灾害场景下的异质影像变化检测难题。该网络构建了基于“SAR影像迁移-文本生成”的跨模态语义引导机制，有效缩小了异质影像间的语义鸿沟；同时设计了伪孪生视觉特征提取分支和多维度变化特征增强模块（MCFEM），通过嵌入模态适配器降低遥感影像的域分布差异，结合时序交叉注意力、多粒度差分及混合相似度投影构建变化特征增强模块，实现对时空上下文信息的高效整合。在两个典型异质数据集上的试验结果表明，MCE-CLIP在F₁值和IoU等核心指标上优于现有主流异质变化检测方法。

关键词: 变化检测, 异质遥感影像, 视觉语言模型, 多模态融合, SAR

Abstract:

Heterogeneous change detection using optical and SAR imagery is of great significance for disaster emergency response and all-weather monitoring. However, the significant differences in their imaging mechanisms lead to inconsistent feature distributions, which, coupled with the lack of annotated samples and textual descriptions, restrict the detection performance of traditional methods and existing deep learning models. To this end, this paper proposes a multi-dimensional change enhancement CLIP change detection network (MCE-CLIP), aiming to tackle the challenges of heterogeneous image change detection in flood disaster scenarios. The network constructs a cross-modal semantic guidance mechanism based on “SAR image transfer-text generation”, effectively narrowing the semantic gap between heterogeneous images. Meanwhile, a pseudo-siamese visual feature extraction branch and a multi-dimensional change feature enhancement module (MCFEM) are designed. By embedding modality adapters, the domain distribution discrepancy of remote sensing images is reduced. Furthermore, the MCFEM is constructed by integrating temporal cross-attention, multi-granularity differencing, and hybrid similarity projection, achieving the efficient integration of spatiotemporal contextual information. Experimental results on two typical heterogeneous datasets demonstrate that MCE-CLIP outperforms existing mainstream heterogeneous change detection methods in core evaluation metrics such as F₁ score and intersection over union.

Key words: change detection, heterogeneous remote sensing imagery, vision-language model, multi-modal fusion, SAR

中图分类号:

P237

尉锐, 李杰, 刘汇慧, 吴美茹, 林镠鹏, 袁强强, 郑莉. 面向洪水灾害的视觉-文本协同表征的异质遥感变化检测方法[J]. 测绘学报, 2026, 55(5): 927-940.

Rui YU, Jie LI, Huihui LIU, Meiru WU, Liupeng LIN, Qiangqiang YUAN, Li ZHENG. Heterogeneous remote sensing change detection based on vision-language collaborative representation for flood disasters[J]. Acta Geodaetica et Cartographica Sinica, 2026, 55(5): 927-940.

导出引用管理器 EndNote|Reference Manager|ProCite|BibTeX|RefWorks

链接本文: http://xb.chinasmp.com/CN/10.11947/j.AGCS.2026.20250521

http://xb.chinasmp.com/CN/Y2026/V55/I5/927

图/表 16

表1

图1

图2

图3

图4

图5

图6

图7

表2

图8

表3

图9

表4

表5

表6

图10

参考文献 25

[1]	张继贤, 顾海燕, 倪欢, 等. 遥感智能变化检测的深度学习方法：演变与发展趋势[J]. 测绘学报, 2025, 54(8): 1347-1370. DOI: . doi: 10.11947/j.AGCS.2025.20240417
	ZHANG Jixian, GU Haiyan, NI Huan, et al. Deep learning methods for remote sensing intelligent change detection: evolution and development[J]. Acta Geodaetica et Cartographica Sinica, 2025, 54(8): 1347-1370. DOI: . doi: 10.11947/j.AGCS.2025.20240417
[2]	龚良雄, 李星华, 程远明, 等. 时空差异增强与自适应特征融合的轻量级遥感影像变化检测网络[J]. 测绘学报, 2025, 54(1): 136-153. DOI: . doi: 10.11947/j.AGCS.2025.20240299
	GONG Liangxiong, LI Xinghua, CHENG Yuanming, et al. A lightweight remote sensing images change detection network utilizing spatio-temporal difference enhancement and adaptive feature fusion[J]. Acta Geodaetica et Cartographica Sinica, 2025, 54(1): 136-153. DOI: . doi: 10.11947/j.AGCS.2025.20240299
[3]	李健慷, 张桂欣, 祝善友, 等. 融合多尺度特征Transformer的高分辨率遥感图像变化检测[J]. 遥感学报, 2025, 29(1): 266-278.
	LI Jiankang, ZHANG Guixin, ZHU Shanyou, et al. Change detection for high-resolution remote sensing images with multi-scale feature transformer[J]. National Remote Sensing Bulletin, 2025, 29(1): 266-278.
[4]	LIU W, JI X, LIU J, et al. A novel unsupervised change detection method with structure consistency and GFLICM based on UAV images[J]. Journal of Geodesy and Geoinformation Science, 2022, 5(1): 91-102.
[5]	杨彬, 毛银, 陈晋, 等. 深度学习的遥感变化检测综述：文献计量与分析[J]. 遥感学报, 2023, 27(9): 1988-2005.
	YANG Bin, MAO Yin, CHEN Jin, et al. Review of remote sensing change detection in deep learning: bibliometric and analysis[J]. National Remote Sensing Bulletin, 2023, 27(9): 1988-2005.
[6]	柳思聪, 都科丞, 郑永杰, 等. 人工智能时代的遥感变化检测技术：继承、发展与挑战[J]. 遥感学报, 2023, 27(9): 1975-1987.
	LIU Sicong, DU Kecheng, ZHENG Yongjie, et al. Remote sensing change detection technology in the era of artificial intelligence: inheritance, development and challenges[J]. Journal of Remote Sensing, 2023, 27(9): 1975-1987.
[7]	成飞飞, 付志涛, 黄亮, 等. 深度学习在光学和SAR影像融合研究进展[J]. 遥感学报, 2022, 26(9): 1744-1756.
	CHENG Feifei, FU Zhitao, HUANG Liang, et al. Review of deep learning in optical and SAR image fusion[J]. National Remote Sensing Bulletin, 2022, 26(9): 1744-1756.
[8]	MA Zongfang, WANG Ruiqi, HAO Fan, et al. Heterogeneous image change detection based on dual image translation and dual contrastive learning[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 4704714.
[9]	汤玉奇, 林泽锋, 韩特, 等. 基于对称网络的光学和SAR影像变化检测[J]. 遥感学报, 2024, 28(6): 1560-1575.
	TANG Yuqi, LIN Zefeng, HAN Te, et al. Optical and SAR image change detection based on a symmetric network[J]. National Remote Sensing Bulletin, 2024, 28(6): 1560-1575.
[10]	LIU Jia, GONG Maoguo, QIN Kai, et al. A deep convolutional coupling network for change detection based on heterogeneous optical and radar images[J]. IEEE Transactions on Neural Networks and Learning Systems, 2018, 29(3): 545-559.
[11]	LI Xinghua, DU Zhengshun, HUANG Yanyuan, et al. A deep translation (GAN) based change detection network for optical and SAR remote sensing images[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2021, 179: 14-34.
[12]	LI Jie, WU Meiru, LIN Liupeng, et al. GLCD-DA: change detection from optical and SAR imagery using a global-local network with diversified attention[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2025, 226: 396-414.
[13]	RADFORD A, KIM J W, HALLACY C, et al. Learning transferable visual models from natural language supervision[C]//Proceedings of 2021 International Conference on Machine Learning. San Diego: IEEE, 2021.
[14]	ELGENDY H, SHARSHAR A, ABOEITTA A, et al. GeoLLaVA: efficient fine-tuned vision-language models for temporal change detection in remote sensing[EB/OL]. [2025-10-12]. https://arxiv.org/abs/2410.19552.
[15]	DONG Sijun, WANG Libo, DU Bo, et al. ChangeCLIP: remote sensing change detection with multimodal vision-language representation learning[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2024, 208: 53-69.
[16]	QIU Junlong, LIU Wei, ZHANG Hui, et al. A novel change detection method based on visual language from high-resolution remote sensing images[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2025, 18: 4554-4567.
[17]	CHEN Runfa, HUANG Wenbing, HUANG Binghui, et al. Reusing discriminators for encoding: towards unsupervised image-to-image translation[C]//Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle: IEEE, 2020: 8165-8174.
[18]	PANG Yingxue, LIN Jianxin, QIN Tao, et al. Image-to-image translation: methods and applications[J]. IEEE Transactions on Multimedia, 2022, 24: 3859-3881.
[19]	ZAN Yujie, JI Shunping, CHAO Songtao, et al. Open-vocabulary generative vision-language models for creating a large-scale remote sensing change detection dataset[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2025, 225: 275-290.
[20]	MIGNOTTE M. A fractal projection and Markovian segmentation-based approach for multimodal change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(11): 8046-8058.
[21]	LUPPINO L T, BIANCHI F M, MOSER G, et al. Unsupervised image regression for heterogeneous change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(12): 9960-9975.
[22]	SUN Yuli, LEI Lin, LI Xiao, et al. Structure consistency-based graph for unsupervised change detection with homogeneous and heterogeneous remote sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 4700221.
[23]	LUPPINO L T, KAMPFFMEYER M, BIANCHI F M, et al. Deep image translation with an affinity-based change prior for unsupervised multimodal change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 4700422.
[24]	CAYE DAUDT R, LE SAUX B, BOULCH A. Fully convolutional Siamese networks for change detection[C]//Proceedings of the 25th IEEE International Conference on Image Processing. Athens: IEEE, 2018: 4063-4067.
[25]	LÜ Zhiyong, HUANG Haitao, SUN Weiwei, et al. Novel enhanced UNet for change detection using multi-modal remote sensing image[J]. IEEE Geoscience and Remote Sensing Letters, 2023, 20: 2505405.

类别分类	基础地物实体	洪水灾害状态描述
对象级	河流、湖泊、池塘、湿地、海洋	河流溢出、水体扩张、浑浊水域、深水区
	森林、农田、草地、裸土、植被、沙漠、山脉、丛林、岛屿	被淹没的农田、植被受损、泥泞土地、浸水区
	道路、桥梁、建筑物、住宅区、商业区、工业区、校园、广场、停车场、不透水地表	被淹没的道路、切断的交通、房屋被围困、城市内涝
场景级	乡村场景、城市景观、沿海地区、沿河地区	大范围淹没、洪水灾害区域、灾后废墟

类型	方法	OA	Pr	Re	F₁值	IoU
无监督	INLPG	88.84	42.90	40.81	41.83	26.44
	SCCN	80.22	29.67	71.98	42.02	26.60
	ACE-Net	91.02	54.04	57.68	55.80	38.70
风格迁移	FC-EF	97.18	90.90	79.93	85.06	74.00
	E-UNet	97.17	85.22	86.20	85.71	74.99
	DTCDN	98.98	94.98	94.89	94.93	90.34
	GLCD-DA	99.20	95.24	96.65	95.94	92.20
VLM	ChangeCLIP	98.65	95.75	90.30	92.94	86.82
	VLCD	98.43	97.84	85.93	91.50	84.33
	MCE-CLIP	99.22	96.23	95.87	96.05	92.84

类型	方法	OA	Pr	Re	F₁值	IoU
无监督	INLPG	91.99	45.66	58.81	51.41	34.60
	SCCN	88.02	39.15	89.85	54.54	37.49
	ACE-Net	92.26	47.31	65.36	54.89	37.82
风格迁移	FC-EF	94.55	70.60	60.42	65.12	48.28
	E-UNet	95.69	70.84	68.55	69.67	53.46
	DTCDN	95.19	64.15	75.09	69.19	52.90
	GLCD-DA	95.76	68.68	75.62	71.98	56.23
VLM	ChangeCLIP	95.65	70.74	67.55	69.11	52.80
	VLCD	95.86	76.06	62.08	68.36	51.93
	MCE-CLIP	96.02	72.56	71.97	72.27	56.58

数据集	指标	GLCD-DA	MCE-CLIP	p值	显著性
Gloucester Ⅰ	F₁值	95.32±0.38	95.98±0.17	p<0.01	显著
Gloucester Ⅰ	IoU	91.05±0.45	92.27±0.21	p<0.01	显著
California	F₁值	71.35±0.42	72.18±0.11	p<0.01	显著
California	IoU	55.45±0.51	56.40±0.15	p<0.01	显著

输入影像类型	核心模块			Gloucester Ⅰ数据集					California数据集
输入影像类型	文本提示	模态适配器	MCFEM	OA	Pr	Re	F₁值	IoU	OA	Pr	Re	F₁值	IoU
光学/SAR	×	×	×	98.69	98.48	88.06	92.98	86.88	94.86	61.27	77.91	68.59	52.20
光学/SAR	√	×	×	98.89	98.00	90.57	94.14	88.93	94.81	60.62	79.86	68.92	52.58
光学/SAR	√	√	×	98.92	97.11	91.76	94.36	89.32	95.90	76.77	61.73	68.44	52.02
光学/SAR	√	×	√	99.14	96.45	94.75	95.59	91.56	95.88	75.35	63.57	68.96	52.63
光学/伪光学	√	√	√	99.18	97.29	94.05	95.79	91.92	95.61	66.76	77.89	71.90	56.13
光学/SAR	√	√	√	99.22	96.23	95.87	96.05	92.84	96.02	72.56	71.97	72.27	56.58

面向洪水灾害的视觉-文本协同表征的异质遥感变化检测方法

Heterogeneous remote sensing change detection based on vision-language collaborative representation for flood disasters

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 16

参考文献 25

相关文章 15

编辑推荐

Metrics

本文评价

K	OA	Pr	Re	F₁值	IoU
3	98.86	96.46	91.83	94.09	88.83
5	99.19	97.59	94.05	95.79	91.92
7	99.22	96.23	95.87	96.05	92.84
9	99.21	97.21	94.64	95.91	92.13
11	99.16	96.28	95.13	95.7	91.76

[1]	王红美, 王利花, 谭本华, 姜晓轶, 宋丽丽, 孙伟伟. 中国近海海表流场SAR高精度反演研究[J]. 测绘学报, 2026, 55(4): 588-603.
[2]	吴岚昕, 彭江涛, 孙伟伟, 杨冰. 面向海岸带湿地高光谱遥感的欧拉映射与互补特征建模变化检测方法[J]. 测绘学报, 2026, 55(4): 618-631.
[3]	路中, 赵金奇, 牛玉芬, 陈立权, 樊茜佑, 司锦钊, 王子璇, 高源, 王帅, 曲菲霏, 时洪涛, 闫世勇, 师芸, 赵争. NISAR卫星对地观测革新及其在地表形变监测中的应用[J]. 测绘学报, 2026, 55(2): 261-274.
[4]	彭代锋, 刘雪莲, 鲁梦飞, 管海燕. 基于多尺度跨模态特征融合的异源遥感影像洪水变化检测[J]. 测绘学报, 2026, 55(2): 328-343.
[5]	胡浩鹏, 吴杭彬, 战仕浩, 温在豪, 刘春. 视觉点云质量优化支持的道路杆状物变化检测[J]. 测绘学报, 2026, 55(2): 344-358.
[6]	韩斌, 黄欣, 李丰毅, 卢晓珍. 一种双编码器自适应特征融合的SAR图像水体分割网络[J]. 测绘学报, 2026, 55(1): 101-113.
[7]	张继贤, 顾海燕, 倪欢, 李海涛, 杨懿, 丁少鹏, 隋淞蔓. 遥感智能变化检测的深度学习方法：演变与发展趋势[J]. 测绘学报, 2025, 54(8): 1347-1370.
[8]	李鹏, 白建博, 李振洪, 王厚杰. 融合多轨道TS-InSAR的广域海岸带地面沉降监测及成因解析——以山东省为例[J]. 测绘学报, 2025, 54(7): 1178-1191.
[9]	王超, 陈天宇, 张同, AhmedTanvir, 纪立强, 谢涛, 杨佳俊, 王帅. 基于全局差分增强模块和平衡惩罚损失的多源光学遥感影像变化检测[J]. 测绘学报, 2025, 54(5): 873-887.
[10]	罗卿莉, 李雪岩, 黄国满, 陈红辉, 薛铭龙, 李健. AOSN：α-最优网络模型的山区单通道SAR高程重建方法[J]. 测绘学报, 2025, 54(5): 888-898.
[11]	邢广澳, 卢官明, 韩斌. MAFUNet：结合注意力机制和主动轮廓损失的SAR图像水体分割算法[J]. 测绘学报, 2025, 54(5): 924-936.
[12]	杜钊锋, 李国鹏, 刘站科, 尚夏明, 康胜军, 王晓强. 国家重点沉降区域多监测手段综合分析[J]. 测绘学报, 2025, 54(3): 481-492.
[13]	施宏凯, 何秀凤, 吴怿昊, 郑翔天, 宋敏峰. 近海强干扰区域高频全聚焦SAR波形污染识别与海面高精确提取算法[J]. 测绘学报, 2025, 54(2): 272-285.
[14]	彭葳, 阳靖, 付海强, 朱建军, 曾东. 顾及运动误差特性的无人机载重轨InSAR数据处理方法[J]. 测绘学报, 2025, 54(11): 1954-1967.
[15]	陈海禄, 沈云中. GNSS辅助下的InSAR对流层延迟垂直分层与湍流分量联合改正模型[J]. 测绘学报, 2025, 54(10): 1786-1797.