Dual-Granularity Feature Alignment for Change Detection in Remote Sensing Images
Deep learning has emerged as the preferred method for remote sensing change detection owing to its ability to automatically extract discriminative features from bitemporal images. However, few methods simultaneously consider heterogeneous appearance of objects and affine geometric difference between...
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10830007/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832542605946126336 |
|---|---|
| author | Feng Zhou Xinyu Zhang Hui Shuai Renlong Hang Shanshan Zhu Tianyu Geng |
| author_facet | Feng Zhou Xinyu Zhang Hui Shuai Renlong Hang Shanshan Zhu Tianyu Geng |
| author_sort | Feng Zhou |
| collection | DOAJ |
| description | Deep learning has emerged as the preferred method for remote sensing change detection owing to its ability to automatically extract discriminative features from bitemporal images. However, few methods simultaneously consider heterogeneous appearance of objects and affine geometric difference between bitemporal images, both of which contribute to pseudochange. In this article, dual-granularity feature alignment (DgFA) is proposed to deal with these two issues. Specifically, bitemporal features extracted by transformer, along with learnable class tokens, are input into the proposed semantic alignment module to adjust the appearance of separate instances from same-category objects to ensure a cohesive style. Then, a spatial alignment module is introduced to use the estimated transformation field to accomplish bitemporal feature registration. Finally, we develop a temporal contrast-based change detection head to infer the change map based on dual-granularity aligned bitemporal features and corresponding difference maps. To refine the change map, this head also constrains the feature similarity within changed and unchanged regions across bitemporal features via a contrastive loss. Experimental results demonstrate that DgFA outperforms several state-of-the-art methods on three public benchmark datasets, including LEVIR-CD, CDD, and SYSU-CD. |
| format | Article |
| id | doaj-art-ba769e754df343d7b847d694fa3098af |
| institution | Kabale University |
| issn | 1939-1404 2151-1535 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj-art-ba769e754df343d7b847d694fa3098af2025-02-04T00:00:15ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-01184487449710.1109/JSTARS.2025.352679510830007Dual-Granularity Feature Alignment for Change Detection in Remote Sensing ImagesFeng Zhou0https://orcid.org/0000-0002-1519-1001Xinyu Zhang1Hui Shuai2https://orcid.org/0000-0001-8840-5069Renlong Hang3https://orcid.org/0000-0001-6046-3689Shanshan Zhu4Tianyu Geng5Jiangsu Modern Intelligent Audit Integrated Application Technology Engineering Research Center, School of Computer Science, Nanjing Audit University, Nanjing, ChinaJiangsu Emergency Early Warning Release Center, Nanjing, ChinaSchool of Computer Science, Nanjing University of Posts and Telecommunications, Nanjing, ChinaJiangsu Collaborative Innovation Center of Atmospheric Environment and Equipment Technology, School of Computer and Software, Nanjing University of Information Science and Technology, Nanjing, ChinaSchool of Computer Science, Nanjing Audit University, Nanjing, ChinaSchool of Computer Science, Nanjing Audit University, Nanjing, ChinaDeep learning has emerged as the preferred method for remote sensing change detection owing to its ability to automatically extract discriminative features from bitemporal images. However, few methods simultaneously consider heterogeneous appearance of objects and affine geometric difference between bitemporal images, both of which contribute to pseudochange. In this article, dual-granularity feature alignment (DgFA) is proposed to deal with these two issues. Specifically, bitemporal features extracted by transformer, along with learnable class tokens, are input into the proposed semantic alignment module to adjust the appearance of separate instances from same-category objects to ensure a cohesive style. Then, a spatial alignment module is introduced to use the estimated transformation field to accomplish bitemporal feature registration. Finally, we develop a temporal contrast-based change detection head to infer the change map based on dual-granularity aligned bitemporal features and corresponding difference maps. To refine the change map, this head also constrains the feature similarity within changed and unchanged regions across bitemporal features via a contrastive loss. Experimental results demonstrate that DgFA outperforms several state-of-the-art methods on three public benchmark datasets, including LEVIR-CD, CDD, and SYSU-CD.https://ieeexplore.ieee.org/document/10830007/Appearance differencechange detectioncontrastive lossfeature alignmentgeometric difference |
| spellingShingle | Feng Zhou Xinyu Zhang Hui Shuai Renlong Hang Shanshan Zhu Tianyu Geng Dual-Granularity Feature Alignment for Change Detection in Remote Sensing Images IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Appearance difference change detection contrastive loss feature alignment geometric difference |
| title | Dual-Granularity Feature Alignment for Change Detection in Remote Sensing Images |
| title_full | Dual-Granularity Feature Alignment for Change Detection in Remote Sensing Images |
| title_fullStr | Dual-Granularity Feature Alignment for Change Detection in Remote Sensing Images |
| title_full_unstemmed | Dual-Granularity Feature Alignment for Change Detection in Remote Sensing Images |
| title_short | Dual-Granularity Feature Alignment for Change Detection in Remote Sensing Images |
| title_sort | dual granularity feature alignment for change detection in remote sensing images |
| topic | Appearance difference change detection contrastive loss feature alignment geometric difference |
| url | https://ieeexplore.ieee.org/document/10830007/ |
| work_keys_str_mv | AT fengzhou dualgranularityfeaturealignmentforchangedetectioninremotesensingimages AT xinyuzhang dualgranularityfeaturealignmentforchangedetectioninremotesensingimages AT huishuai dualgranularityfeaturealignmentforchangedetectioninremotesensingimages AT renlonghang dualgranularityfeaturealignmentforchangedetectioninremotesensingimages AT shanshanzhu dualgranularityfeaturealignmentforchangedetectioninremotesensingimages AT tianyugeng dualgranularityfeaturealignmentforchangedetectioninremotesensingimages |