A lightweight remote sensing images change detection network utilizing spatio-temporal difference enhancement and adaptive feature fusion

doi:10.11947/j.AGCS.2025.20240299

Abstract

Abstract:

To address the limitations in existing change detection methods of remote sensing images, such as insufficient utilization of multi-temporal difference features and inadequate multi-scale feature fusion, a lightweight remote sensing images change detection network named SEAFNet is proposed, which integrates spatio-temporal difference enhancement with adaptive feature fusion. This paper designs the lightweight spatio-temporal difference enhancement module, which employs a dual-branch structure with semantic change perception and spatial change perception. This module combines a semantic adaptive enhancement mechanism and a mixed attention mechanism to enhance the space-spectrum differences in the bi-temporal feature maps. To further refine the edges of the change regions, different scale feature maps are optimized through an edge refinement residual module. The bi-directional feature fusion pyramid structure is also improved by using learnable weight parameters to quantify the importance of features at different scales, achieving effective multi-scale feature fusion. Comparative experiments with ten mainstream change detection methods on WHU-CD, LEVIR-CD, SYSU-CD and SECOND datasets demonstrate that SEAFNet outperforms these methods in qualitative and quantitative analysis, and the balance between network complexity and accuracy.

Key words: remote sensing images, spatio-temporal difference enhancement, attention mechanism, adaptive feature fusion, change detection

CLC Number:

P237

Liangxiong GONG, Xinghua LI, Yuanming CHENG, Xingyou ZHAO, Renping XIE, Honggen WANG. 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.

Figures/Tables 20

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Tab. 1

Tab. 2

Fig. 8

Tab. 3

Fig. 9

Fig. 10

Fig. 11

Fig. 12

Fig. 13

Tab. 4

Tab. 5

Tab. 6

Fig. 14

References 50

[1]	刘宣广, 李蒙蒙, 汪小钦, 等. 基于面向对象孪生神经网络的高分辨率遥感影像建筑物变化检测[J]. 遥感学报, 2024, 28(2): 437-454.
	LIU Xuanguang, LI Mengmeng, WANG Xiaoqin, et al. Use of object-based siamese neural network to build change detection from very high resolution remote-sensing images[J]. National Remote Sensing Bulletin, 2024, 28(2): 437-454.
[2]	王超, 王帅, 陈晓, 等. 联合UNet++和多级差分模块的多源光学遥感影像对象级变化检测[J]. 测绘学报, 2023, 52(2): 283-296. DOI:. doi: 10.11947/j.AGCS.2023.20220202
	WANG Chao, WANG Shuai, CHEN Xiao, et al. Object-level change detection of multi-sensor optical remote sensing images combined with UNet++ and multi-level difference module[J]. Acta Geodaetica et Cartographica Sinica, 2023, 52(2): 283-296. DOI:. doi: 10.11947/j.AGCS.2023.20220202
[3]	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]. National Remote Sensing Bulletin, 2023, 27(9): 1975-1987.
[4]	徐炜, 王驿飞, 张艳, 等. 基于双边滤波和小目标抑制的异源遥感变化检测[J]. 电子测量技术, 2021, 44(17): 165-172.
	XU Wei, WANG Yifei, ZHANG Yan, et al. Heterogeneous remote sensing image change detection based on bilateral filtering and small target suppression[J]. Electronic Measurement Technology, 2021, 44(17): 165-172.
[5]	邱云飞, 屈照阳, 方立. 结合差异增强的异源遥感影像变化检测网络[J]. 测绘科学, 2024, 49(1): 153-162.
	QIU Yunfei, QU Zhaoyang, FANG Li. Change detection network for heterogeneous remote sensing images with differential augmentation[J]. Science of Surveying and Mapping, 2024, 49(1): 153-162.
[6]	张玉沙, 黄岩, 谭琨, 等. 基于多分类器集成和对象的城市典型地物要素变化检测——以ZY-3影像为例[J]. 地理与地理信息科学, 2018, 34(3): 54-60,130.
	ZHANG Yusha, HUANG Yan, TAN Kun, et al. Change detection of urban typical features based on multi-classifier ensemble and objeets: taking ZY-3 as an example[J]. 2018, 34(3): 54-60,130.
[7]	张晗, 倪维平, 严卫东, 等. 利用分形和多尺度分析的中低分辨率SAR图像变化检测[J]. 武汉大学学报(信息科学版), 2016, 41(5): 642-648.
	ZHANG Han, NI Weiping, YAN Weidong, et al. Mid and low resolution SAR image change detection based on fractal and multi-scale analysis[J]. Geomatics and Information Science of Wuhan University, 2016, 41(5): 642-648.
[8]	赵元昊, 孙根云, 张爱竹, 等. 基于典型相关分析的遥感影像非监督超像素级变化检测[J]. 遥感学报, 2024, 28(4): 1025-1040.
	ZHAO Yuanhao, SUN Genyun, ZHANG Aizhu, et al. Unsupervised super pixel level change detection based on canonical correlation analysis[J]. National Remote Sensing Bulletin, 2024, 28(4): 1025-1040.
[9]	ZHOU Zhenjin, MA Lei, FU Tengyun, et al. Change detection in coral reef environment using high-resolution images: comparison of object-based and pixel-based paradigms[J]. International Journal of Geo-Information, 2018, 7(11): 441.
[10]	胡永月, 肖鹏峰, 冯学智, 等. 面向对象变化检测中多时相图像分割模式影响评价[J]. 南京大学学报(自然科学), 2015, 51(5): 1049-1057.
	HU Yongyue, XIAO Pengfeng, FENG Xuezhi, et al. Evaluating the effectiveness of multi-temporal image segmentation on object-based change detection[J]. Journal of Nanjing University (Natural Sciences), 2015, 51(5): 1049-1057.
[11]	韩潇冰. 高分辨率遥感影像“像素-目标-场景”的深度理解方法研究[D]. 武汉: 武汉大学, 2020.
	HAN Xiaobing. “Pixel-obiect-scene” level deep understanding for high spatial resolution remote sensing imagery[D]. Wuhan: Wuhan University, 2020.
[12]	WU Chen, ZHANG Liangpei, DU Bo. Kernel slow feature analysis for scene change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(4): 2367-2384.
[13]	CHEN Jie, HUANG Haozhe, PENG Jian, et al. Contextual information-preserved architecture learning for remote sensing scene classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 1-14.
[14]	WANG Min, WANG Peidong. CFM-UNet: a joint CNN and transformer network via cross feature modulation for remote sensing images segmentation[J]. Journal of Geodesy and Geoinformation Science, 2023, 6(4): 40-47.
[15]	ZHU Qiqi, GUO Xi, DENG Weihuan, et al. Land-use/land-cover change detection based on a siamese global learning framework for high spatial resolution remote sensing imagery[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2022, 184: 63-78.
[16]	CHEN Zhanlong, LI Shuangjiang, XU Yongyang, et al. Correg-Yolov3: a method for dense buildings detection in high-resolution remote sensing images[J]. Journal of Geodesy and Geoinformation Science, 2023, 6(2): 51-61.
[17]	CODEGONI A, LOMBARDI G, FERRARI A. Tinycd: a (not so) deep learning model for change detection[J]. Neural Computing and Applications, 2023, 35(11): 8471-8486.
[18]	YANG Haiping, CHEN Yuanyuan, WU Wei, et al. A lightweight siamese neural network for building change detection using remote sensing images[J]. Remote Sensing, 2023, 15(4): 928.
[19]	YOU Zhihui, CHEN Sibao, WANG Jiaxin, et al. Robust feature aggregation network for lightweight and effective remote sensing image change detection[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2024, 215: 31-43.
[20]	HUANG Yanyuan, LI Xinghua, DU Zhengshun, et al. Spatiotemporal enhancement and interlevel fusion network for remote sensing images change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 1-14.
[21]	BANDARA W G C, PATEL V M. A transformer-based siamese network for change detection[C]//Proceedings of 2022 IEEE International Geoscience and Remote Sensing Symposium. Malaysia: IEEE, 2022: 207-210.
[22]	ZHANG Cui, WANG Liejun, CHENG Shuli, et al. SwinSUNet: pure transformer network for remote sensing image change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 1-13.
[23]	WANG Decheng, CHEN Xiangning, JIANG Mingyong, et al. ADS-Net: an attention-based deeply supervised network for remote sensing image change detection[J]. International Journal of Applied Earth Observation and Geoinformation, 2021, 101: 102348.
[24]	张建兵, 严泽枭, 马淑芳. 用于遥感影像建筑物变化检测的多尺度交叉对偶注意力网络[J]. 地球信息科学学报, 2023, 25(12): 2487-2500.
	ZHANG Jianbing, YAN Zexiao, MA Shufang. Multi-scale cross dual attention network for building change detection in remote sensing images[J]. Geo-Information Science, 2023, 25(12): 2487-2500.
[25]	LI Qingyang, ZHONG Ruofei, DU Xin, et al. TransUNetCD: a hybrid transformer network for change detection in optical remote-sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 1-9.
[26]	LIANG Shike, HUA Zhen, LI Jinjiang. Enhanced self-attention network for remote sensing building change detection[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2023, 16: 4900-4915.
[27]	CHEN Hongruixuan, SONG Jian, HAN Chengxi, et al. ChangeMamba: remote sensing change detection with spatiotemporal state space model[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 1-20.
[28]	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.
[29]	LI Shujun, HUO Lianzhi. Remote sensing image change detection based on fully convolutional network with pyramid attention[C]//Proceedings of 2021 IEEE International Geoscience and Remote Sensing Symposium. Brussels: IEEE, 2021: 4352-4355.
[30]	姜明, 张新长, 孙颖, 等. 全尺度特征聚合的高分辨率遥感影像变化检测网络[J]. 测绘学报, 2023, 52(10): 1738-1748. DOI:. doi: 10.11947/j.AGCS.2023.20220505
	JIANG Ming, ZHANG Xinchang, SUN Ying, et al. Full-scale feature aggregation network for high-resolution remote sensing image change detection[J]. Acta Geodaetica et Cartographica Sinica, 2023, 52(10): 1738-1748. DOI:. doi: 10.11947/j.AGCS.2023.20220505
[31]	HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 770-778.
[32]	HOWARD A, SANDLER M, CHEN Bo, et al. Searching for MobileNetV3[C]//Proceedings of 2019 IEEE/CVF International Conference on Computer Vision. Seoul: IEEE, 2019: 1314-1324.
[33]	LIU Ze, LIN Yutong, CAO Yue, et al. Swin transformer: hierarchical vision transformer using shifted windows[C]//Proceedings of 2021 IEEE/CVF International Conference on Computer Vision. Montreal: IEEE, 2021: 9992-10002.
[34]	WANG Ao, CHEN Hui, Lin Zijia, et al. RepViT: revisiting mobile CNN from ViT perspective[C]//Proceedings of 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle: IEEE, 2024: 15909-15920.
[35]	FU Jun, LIU Jing, JIANG Jie, et al. Scene segmentation with dual relation-aware attention network[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 32(6): 2547-2560.
[36]	WANG Qilong, WU Banggu, ZHU Pengfei, et al. ECA-Net: efficient channel attention for deep convolutional neural networks[C]//Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle: IEEE, 2020: 11531-11539.
[37]	LIN T Y, DOLLAR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 936-944.
[38]	LIU Shu, QI Lu, QIN Haifang, et al. Path aggregation network for instance segmentation[C]//Proceedings of 2018 IEEE Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018: 8759-8768.
[39]	JI Shunping, WEI Shiqing, LU Meng. Fully convolutional networks for multisource building extraction from an open aerial and satellite imagery data set[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(1): 574-586.
[40]	CHEN Hao, SHI Zhenwei. A spatial-temporal attention-based method and a new dataset for remote sensing image change detection[J]. Remote Sensing, 2020, 12(10): 1662.
[41]	SHI Qian, LIU Mengxi, LI Shengchen, et al. A deeply supervised attention metric-based network and an open aerial image dataset for remote sensing change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 1-16.
[42]	YANG Kunping, XIA Guisong, LIU Zicheng, et al. A symmetric siamese networks for semantic change detection in aerial images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 1-18.
[43]	LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[J]. IEEE Transactions on Pattern Analysis & AMP, Machine Intelligence, 2020, 42(2): 318-327.
[44]	SUDRE H C, LI W Q, TOM V, et al. Generalised dice overlap as a deep learning loss function for highly unbalanced segmentations[C]//Proceedings of 2017 Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support. Quebec City: Springer, 2017: 240-248.
[45]	FANG Sheng, LI Kaiyu, SHAO Jinyuan, et al. SNUNet-CD: a densely connected siamese network for change detection of VHR images[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19: 1-5.
[46]	CHEN Hao, QI Zipeng, SHI Zhenwei. Remote sensing image change detection with transformers[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 1-14.
[47]	LIU Wei, LIN Yiyuan, LIU Weijia, et al. An attention-based multiscale transformer network for remote sensing image change detection[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2023, 202: 599-609.
[48]	LI Zhenglai, TANG Chang, WANG Lizhe. Remote sensing change detection via temporal feature interaction and guided refinement[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 1-11.
[49]	FENG Yuchao, JIANG Jiawei, XU Honghui, et al. Change detection on remote sensing images using dual-branch multilevel intertemporal network[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 1-5.
[50]	WEI Jinjiang, SUN Kaimin, LI Wenzhuo, et al. Robust change detection for remote sensing images based on temporospatial interactive attention module[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2024, 128: 103767.

主干网络	主干网络复杂度嵌入后网络复杂度				WHU-CD/(%)			LEVIR-CD/(%)			SYSU-CD/(%)
主干网络	参数量/M	FLOPs/G	参数量/M	FLOPs/G	IoU	F₁值	OA	IoU	F₁值	OA	IoU	F₁值	OA
Res Net18	11.18	19.05	13.44	22.31	77.12	87.08	98.83	78.61	88.02	98.85	67.23	80.40	91.23
Res Net34	21.28	38.43	23.55	41.69	72.08	83.78	98.49	79.53	88.60	98.89	66.69	80.01	91.21
V3-Small	1.14	1.35	1.73	2.71	78.34	87.86	98.92	76.93	86.96	98.75	65.92	79.46	90.99
V3-Large	3.41	5.79	5.67	9.05	80.59	89.25	99.05	81.23	89.65	98.96	69.46	81.97	91.78
Swin-S	48.76	89.13	32.51	51.28	83.04	90.73	99.22	82.96	91.07	99.11	68.27	81.14	91.48
Swin-B	86.64	158.3	95.52	167.01	80.50	89.20	99.04	81.39	89.74	98.97	65.48	79.14	90.52
Repvit_m1_1	7.77	14.37	10.04	17.63	79.44	88.54	98.99	80.33	89.09	98.92	67.27	80.43	91.26
Repvit_m1_5	13.62	24.43	15.88	27.69	75.54	86.06	98.77	78.22	87.78	98.83	64.70	78.57	90.69

损失函数	WHU-CD			LEVIR-CD			SYSU-CD			SECOND
损失函数	IoU	F₁值	OA	IoU	F₁值	OA	IoU	F₁值	OA	IoU	F₁值	OA
BCE loss+Dice loss	78.88	88.19	98.93	80.74	89.35	98.95	69.55	82.04	91.79	58.38	73.72	94.69
Focal loss+Dice loss	80.59	89.25	99.05	81.23	89.65	98.96	69.46	81.97	91.78	58.93	74.16	94.67

网络模型	网络复杂度		WHU-CD/(%)			LEVIR-CD/(%)			SYSU-CD/(%)			SECOND/(%)
网络模型	参数量/M	FLOPs/G	IoU	F₁值	OA	IoU	F₁值	OA	IoU	F₁值	OA	IoU	F₁值	OA
Tiny-CD	0.29	6.18	66.34	79.76	97.91	82.16	90.21	99.02	65.10	78.86	90.31	57.20	72.78	93.95
RFANet	2.86	12.65	73.28	84.58	98.52	81.11	89.57	98.95	65.61	79.24	90.53	56.51	72.22	94.39
FC-Siam-diff	1.35	18.91	64.05	78.09	97.73	73.41	84.66	98.49	65.35	79.04	90.79	50.42	67.04	93.44
SNUNet	12.03	219.33	71.59	83.45	98.40	81.33	89.49	98.96	65.64	79.25	90.71	55.04	71.00	94.00
AMTNet	16.45	58.85	72.93	84.35	98.56	79.29	88.45	98.84	60.89	75.69	88.91	51.08	67.62	93.65
ChangeFormer	29.75	84.73	70.05	82.39	98.28	79.80	88.76	98.87	63.90	77.97	89.40	51.10	67.64	93.20
BIT	11.47	105.24	72.49	84.11	98.45	80.62	89.27	98.92	63.72	77.30	89.18	56.79	72.44	94.45
TFI-GR	27.78	38.96	76.65	86.70	98.78	80.66	89.29	98.96	67.84	81.00	91.34	55.42	71.32	93.99
CDNeXt	39.42	64.33	76.53	86.53	98.73	80.77	89.36	98.94	67.14	80.34	91.16	56.50	72.20	94.29
DMINet	6.24	59.49	78.04	87.76	98.93	80.25	89.04	98.92	67.72	80.75	91.32	57.68	73.16	94.65
SEAFNet	5.67	9.05	80.77	89.36	99.06	82.74	90.73	99.08	70.64	82.76	92.16	59.79	74.83	94.76

分支名称		WHU-CD/(%)			LEVIR-CD/(%)			SYSU-CD/(%)			SECOND/(%)			网络复杂度
分支1	分支2	IoU	F₁值	OA	IoU	F₁值	OA	IoU	F₁值	OA	IoU	F₁值	OA	参数量/M	FLOPs/G
×	×	74.77	85.56	98.77	71.25	83.21	98.49	60.55	75.43	89.92	49.91	66.59	93.42	3.57	8.79
×	√	76.35	86.59	98.76	73.58	84.78	98.59	64.69	78.56	90.67	50.83	67.40	93.82	3.55	8.77
√	×	76.04	86.39	98.88	73.98	85.04	98.45	63.60	77.75	90.49	51.34	67.84	93.49	5.56	9.06
√	√	77.60	87.39	98.93	77.76	87.49	98.72	65.31	79.01	90.69	52.45	68.81	93.69	5.56	9.06

权重参数	WHU-CD/(%)			LEVIR-CD/(%)			SYSU-CD/(%)			SECOND/(%)			网络复杂度
权重参数	IoU	F₁值	OA	IoU	F₁值	OA	IoU	F₁值	OA	IoU	F₁值	OA	参数量/M	FLOPs/G
×	76.69	86.81	98.94	77.29	87.17	98.73	66.40	79.81	91.35	53.88	70.03	94.21	3.67	8.72
√	77.29	87.19	98.90	77.78	87.52	98.74	66.88	80.15	91.31	54.61	70.64	94.28	3.67	8.72