An end-to-end homography estimation method for large baseline scenes with an attention mechanism
Aiming at the problems of low accuracy and limited adaptability to large baseline scenes and motion blur scenarios in current homography estimation methods, an end-to-end homography estimation methodwith attention mechanism for large baseline scenes was constructed, which utilized unsupervised learn...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Hebei University of Science and Technology
2024-12-01
|
| Series: | Journal of Hebei University of Science and Technology |
| Subjects: | |
| Online Access: | https://xuebao.hebust.edu.cn/hbkjdx/article/pdf/b202406012?st=article_issue |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841559976094466048 |
|---|---|
| author | Zuojiang LIN Xu CAO Wei CHEN Yu BAI Bo MI Xuewei ZHANG Hao FANG |
| author_facet | Zuojiang LIN Xu CAO Wei CHEN Yu BAI Bo MI Xuewei ZHANG Hao FANG |
| author_sort | Zuojiang LIN |
| collection | DOAJ |
| description | Aiming at the problems of low accuracy and limited adaptability to large baseline scenes and motion blur scenarios in current homography estimation methods, an end-to-end homography estimation methodwith attention mechanism for large baseline scenes was constructed, which utilized unsupervised learning for homography estimation. Firstly, by introducing the SE channel attention module, a homography regression network layer with attention mechanism was constructed, enabling the network to learn the inter-channel correlations of images. Secondly, a binary unsupervised loss construction method based on mask and perceptual loss metrics was proposed to enhance the network′s perception range and adaptability to large baseline scenes. Finally, a Homo-COCO synthetic dataset was created, and data augmentation was used to improve the network model′s robustness to changes in lighting and motion blur, resulting in stronger generalization capabilities in real-world scenes. Extensive comparative and ablation experiments demonstrate that this method outperforms existing methods in terms of accuracy and scene adaptability, showing good precision and adaptability. It can effectively estimate image homography and provide accurate parameter estimation for subsequent computer vision tasks such as image stitching and image correction. |
| format | Article |
| id | doaj-art-48320d113f584dd98cbbac1b2230a955 |
| institution | Kabale University |
| issn | 1008-1542 |
| language | zho |
| publishDate | 2024-12-01 |
| publisher | Hebei University of Science and Technology |
| record_format | Article |
| series | Journal of Hebei University of Science and Technology |
| spelling | doaj-art-48320d113f584dd98cbbac1b2230a9552025-01-05T06:35:23ZzhoHebei University of Science and TechnologyJournal of Hebei University of Science and Technology1008-15422024-12-0145666968210.7535/hbkd.2024yx06012b202406012An end-to-end homography estimation method for large baseline scenes with an attention mechanismZuojiang LIN0Xu CAO1Wei CHEN2Yu BAI3Bo MI4Xuewei ZHANG5Hao FANG6China Construction First Group Construction & Development Corporation Limited, Beijing 100102,ChinaSchool of Automation, Beijing Institute of Technology, Beijing 100081, ChinaChina Construction First Group Construction & Development Corporation Limited, Beijing 100102,ChinaChina Construction Municipal Engineering Corporation Limited, Beijing 102627, ChinaChina Construction First Group Construction & Development Corporation Limited, Beijing 100102,ChinaChina Construction Municipal Engineering Corporation Limited, Beijing 102627, ChinaSchool of Automation, Beijing Institute of Technology, Beijing 100081, ChinaAiming at the problems of low accuracy and limited adaptability to large baseline scenes and motion blur scenarios in current homography estimation methods, an end-to-end homography estimation methodwith attention mechanism for large baseline scenes was constructed, which utilized unsupervised learning for homography estimation. Firstly, by introducing the SE channel attention module, a homography regression network layer with attention mechanism was constructed, enabling the network to learn the inter-channel correlations of images. Secondly, a binary unsupervised loss construction method based on mask and perceptual loss metrics was proposed to enhance the network′s perception range and adaptability to large baseline scenes. Finally, a Homo-COCO synthetic dataset was created, and data augmentation was used to improve the network model′s robustness to changes in lighting and motion blur, resulting in stronger generalization capabilities in real-world scenes. Extensive comparative and ablation experiments demonstrate that this method outperforms existing methods in terms of accuracy and scene adaptability, showing good precision and adaptability. It can effectively estimate image homography and provide accurate parameter estimation for subsequent computer vision tasks such as image stitching and image correction.https://xuebao.hebust.edu.cn/hbkjdx/article/pdf/b202406012?st=article_issuecomputer image processing; homography estimation; unsupervised learning; attention mechanism; data augmentation; deep learning |
| spellingShingle | Zuojiang LIN Xu CAO Wei CHEN Yu BAI Bo MI Xuewei ZHANG Hao FANG An end-to-end homography estimation method for large baseline scenes with an attention mechanism Journal of Hebei University of Science and Technology computer image processing; homography estimation; unsupervised learning; attention mechanism; data augmentation; deep learning |
| title | An end-to-end homography estimation method for large baseline scenes with an attention mechanism |
| title_full | An end-to-end homography estimation method for large baseline scenes with an attention mechanism |
| title_fullStr | An end-to-end homography estimation method for large baseline scenes with an attention mechanism |
| title_full_unstemmed | An end-to-end homography estimation method for large baseline scenes with an attention mechanism |
| title_short | An end-to-end homography estimation method for large baseline scenes with an attention mechanism |
| title_sort | end to end homography estimation method for large baseline scenes with an attention mechanism |
| topic | computer image processing; homography estimation; unsupervised learning; attention mechanism; data augmentation; deep learning |
| url | https://xuebao.hebust.edu.cn/hbkjdx/article/pdf/b202406012?st=article_issue |
| work_keys_str_mv | AT zuojianglin anendtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT xucao anendtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT weichen anendtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT yubai anendtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT bomi anendtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT xueweizhang anendtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT haofang anendtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT zuojianglin endtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT xucao endtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT weichen endtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT yubai endtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT bomi endtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT xueweizhang endtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism AT haofang endtoendhomographyestimationmethodforlargebaselinesceneswithanattentionmechanism |