基于注意力引导的多尺度特征融合的遥感图像变化检测算法研究
Research on Change Detection Algorithm for Remote Sensing Images Based on the Attention-Guided Multiscale Feature Fusion
摘要: 为了解决遥感图像变化检测中变化边缘检测不清晰以及极小目标的漏检问题，本论文提出了一种基于注意力引导的多尺度特征融合的遥感图像变化检测网络(MFM-CDNet)，旨在提高变化检测的效率和准确性。MFM-CDNet通过特征增强模块FEM和注意力引导的特征融合模块FFM，实现了从粗到细的多尺度特征提取与融合。FEM利用双重注意力机制精确定位图像中的重要区域，而FFM通过对多尺度特征的有效整合与优化，增强了模型对正样本的识别能力，并提高了边缘变化区域的定位精度。在LEVIR-CD、WHU-CD和GoogleGZ-CD三个公开数据集上的实验表明，MFM-CDNet网络综合性能优于现有的变化检测方法。消融实验进一步验证了FEM和FFM模块在提升检测性能中的关键作用。MFM-CDNet的成功应用展示了其在复杂遥感图像变化检测中的鲁棒性和泛化能力，为未来的遥感图像变化检测研究提供了重要的参考。
Abstract: In order to solve the problems of unclear detection of change edges and missed detection of very small targets in remote sensing image change detection, this thesis proposes an attention-guided multi-scale feature fusion-based remote sensing image change detection network (MFM-CDNet), which aims to improve the efficiency and accuracy of the change detection. MFM-CDNet achieves multi-scale feature extraction and fusion from coarse to fine by means of a feature enhancement module, FEM, and an attention-guided feature fusion module, FFM. FFM achieves multi-scale feature extraction and fusion from coarse to fine. FEM utilises a dual attention mechanism to pinpoint important regions in the image, while FFM enhances the model’s ability to identify positive samples and improves the localisation accuracy of edge-change regions through the effective integration and optimisation of multi-scale features. Experiments on three publicly available datasets, LEVIR-CD, WHU-CD and GoogleGZ-CD, show that the comprehensive performance of the MFM-CDNet network outperforms existing change detection methods. The ablation experiments further validate the key role of FEM and FFM modules in enhancing the detection performance. The successful application of MFM-CDNet demonstrates its robustness and generalisation ability in complex remote sensing image change detection, which provides an important reference for future research on remote sensing image change detection.
文章引用：贾正正, 宋媛萌, 张铭钰, 张豪, 韩卓航. 基于注意力引导的多尺度特征融合的遥感图像变化检测算法研究[J]. 计算机科学与应用, 2025, 15(2): 1-12. https://doi.org/10.12677/csa.2025.152028

参考文献

[1] Tison, C., Nicolas, J.-M., Tupin, F. and Maitre, H. (2004) A New Statistical Model for Markovian Classification of Urban Areas in High-Resolution SAR Images. IEEE Transactions on Geoscience and Remote Sensing, 42, 2046-2057. [Google Scholar] [CrossRef] 
[2] Stahl, B., Apfelbeck, J. and Lange, R. (2023) Classification of Micromobility Vehicles in Thermal-Infrared Images Based on Combined Image and Contour Features Using Neuromorphic Processing. Applied Sciences, 13, Article 3795. [Google Scholar] [CrossRef] 
[3] Lee, C., Park, S., Kim, T., Liu, S., Md Reba, M.N., Oh, J., et al. (2022) Machine Learning-Based Forest Burned Area Detection with Various Input Variables: A Case Study of South Korea. Applied Sciences, 12, Article 10077. [Google Scholar] [CrossRef] 
[4] Chen, Y., Duan, Y., Zhang, W., Wang, C., Yu, Q. and Wang, X. (2024) Crop Land Change Detection with MC&N-PSPNet. Applied Sciences, 14, Article 5429. [Google Scholar] [CrossRef] 
[5] Luo, H., Ma, Z., Wu, H., Li, Y., Liu, B., Li, Y., et al. (2023) Validation Analysis of Drought Monitoring Based on FY-4 Satellite. Applied Sciences, 13, Article 9122. [Google Scholar] [CrossRef] 
[6] Liu, M., Liu, J. and Hu, H. (2024) A Novel Deep Learning Network Model for Extracting Lake Water Bodies from Remote Sensing Images. Applied Sciences, 14, Article 1344. [Google Scholar] [CrossRef] 
[7] Affandi, E., Ng, T.F., Pereira, J.J., Ahmad, F. and Banks, V.J. (2023) Revalidation Technique on Landslide Susceptibility Modelling: An Approach to Local Level Disaster Risk Management in Kuala Lumpur, Malaysia. Applied Sciences, 13, Article 768. [Google Scholar] [CrossRef] 
[8] Huang, L., Yao, C., Zhang, L., Luo, S., Ying, F. and Ying, W. (2024) Enhancing Computer Image Recognition with Improved Image Algorithms. Scientific Reports, 14, Article No. 13709. [Google Scholar] [CrossRef] [PubMed]
[9] Wang, M. and Liu, Z. (2024) Defense against Adversarial Attacks in Image Recognition Based on Multilayer Filters. Applied Sciences, 14, Article 8119. [Google Scholar] [CrossRef] 
[10] Huang, L., Chen, Y. and He, X. (2024) Spectral-Spatial Mamba for Hyperspectral Image Classification. Remote Sensing, 16, Article 2449. [Google Scholar] [CrossRef] 
[11] Wang, M., Sun, Y., Xiang, J., Sun, R. and Zhong, Y. (2024) Adaptive Learnable Spectral-Spatial Fusion Transformer for Hyperspectral Image Classification. Remote Sensing, 16, Article 1912. [Google Scholar] [CrossRef] 
[12] Zhang, L., Zhang, L. and Du, B. (2016) Deep Learning for Remote Sensing Data: A Technical Tutorial on the State of the Art. IEEE Geoscience and Remote Sensing Magazine, 4, 22-40. [Google Scholar] [CrossRef] 
[13] Simonyan, K. (2014) Very Deep Convolutional Networks for Large-Scale Image Recognition. arXiv: 1409.1556. [Google Scholar] [CrossRef] 
[14] He, K., Zhang, X., Ren, S. and Sun, J. (2016) Deep Residual Learning for Image Recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, 27-30 June 2016, 770-778. [Google Scholar] [CrossRef] 
[15] Shelhamer, E., Long, J. and Darrell, T. (2017) Fully Convolutional Networks for Semantic Segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39, 640-651. [Google Scholar] [CrossRef] [PubMed]
[16] Lee, J., Wiratama, W., Lee, W., Marzuki, I. and Sim, D. (2023) Bilateral Attention U-Net with Dissimilarity Attention Gate for Change Detection on Remote Sensing Imageries. Applied Sciences, 13, Article 2485. [Google Scholar] [CrossRef] 
[17] Guo, Q., Feng, W., Zhou, C., Huang, R., Wan, L. and Wang, S. (2017) Learning Dynamic Siamese Network for Visual Object Tracking. 2017 IEEE International Conference on Computer Vision (ICCV), Venice, 22-29 October 2017, 1781-1789. [Google Scholar] [CrossRef] 
[18] Zhai, C., Wang, L. and Yuan, J. (2023) New Fusion Network with Dual-Branch Encoder and Triple-Branch Decoder for Remote Sensing Image Change Detection. Applied Sciences, 13, Article 6167. [Google Scholar] [CrossRef] 
[19] Yu, S., Tao, C., Zhang, G., Xuan, Y. and Wang, X. (2024) Remote Sensing Image Change Detection Based on Deep Learning: Multi-Level Feature Cross-Fusion with 3D-Convolutional Neural Networks. Applied Sciences, 14, Article 6269. [Google Scholar] [CrossRef] 
[20] Caye Daudt, R., Le Saux, B. and Boulch, A. (2018) Fully Convolutional Siamese Networks for Change Detection. 2018 25th IEEE International Conference on Image Processing (ICIP), Athens, 7-10 October 2018, 4063-4067. [Google Scholar] [CrossRef] 
[21] Fang, S., Li, K., Shao, J. and Li, Z. (2022) SNUNet-CD: A Densely Connected Siamese Network for Change Detection of VHR Images. IEEE Geoscience and Remote Sensing Letters, 19, 1-5. [Google Scholar] [CrossRef] 
[22] Chen, H., Qi, Z. and Shi, Z. (2022) Remote Sensing Image Change Detection with Transformers. IEEE Transactions on Geoscience and Remote Sensing, 60, 1-14. [Google Scholar] [CrossRef]