Hierarchical Motion Field Alignment for Robust Optical Flow Estimation

Optical flow estimation is a fundamental and long-standing task in computer vision, facilitating the understanding of motion within visual scenes. In this study, we aim to improve optical flow estimation, particularly in challenging scenarios involving small and fast-moving objects. Specifically, we...

Full description

Saved in:
Bibliographic Details
Main Authors: Dianbo Ma, Kousuke Imamura, Ziyan Gao, Xiangjie Wang, Satoshi Yamane
Format: Article
Language:English
Published: MDPI AG 2025-04-01
Series:Sensors
Subjects:
Online Access:https://www.mdpi.com/1424-8220/25/9/2653
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850136425927802880
author Dianbo Ma
Kousuke Imamura
Ziyan Gao
Xiangjie Wang
Satoshi Yamane
author_facet Dianbo Ma
Kousuke Imamura
Ziyan Gao
Xiangjie Wang
Satoshi Yamane
author_sort Dianbo Ma
collection DOAJ
description Optical flow estimation is a fundamental and long-standing task in computer vision, facilitating the understanding of motion within visual scenes. In this study, we aim to improve optical flow estimation, particularly in challenging scenarios involving small and fast-moving objects. Specifically, we proposed a learning-based model incorporating two key components: the Hierarchical Motion Field Alignment module, which ensures accurate estimation of objects of varying sizes while maintaining manageable computational complexity, and the Correlation Self-Attention module, which effectively handles large displacements, making the model suitable for scenarios with fast-moving objects. Additionally, we introduced a Multi-Scale Correlation Search layer to enhance the four-dimensional cost volume, enabling the model to address various types of motion. Experimental results demonstrate that our model achieves superior generalization performance and significantly improves the estimation of small, fast-moving objects.
format Article
id doaj-art-c5b53d0eeccb472fa05225f2a2c45b72
institution OA Journals
issn 1424-8220
language English
publishDate 2025-04-01
publisher MDPI AG
record_format Article
series Sensors
spelling doaj-art-c5b53d0eeccb472fa05225f2a2c45b722025-08-20T02:31:08ZengMDPI AGSensors1424-82202025-04-01259265310.3390/s25092653Hierarchical Motion Field Alignment for Robust Optical Flow EstimationDianbo Ma0Kousuke Imamura1Ziyan Gao2Xiangjie Wang3Satoshi Yamane4Graduate School of Natural Science and Technology, Kanazawa University, Kanazawa 9201192, JapanGraduate School of Natural Science and Technology, Kanazawa University, Kanazawa 9201192, JapanSchool of Information Science, Japan Advanced Institute of Science and Technology, Nomi 9231292, JapanState Key Laboratory of Automotive Simulation and Control, Jilin University, Changchun 130022, ChinaGraduate School of Natural Science and Technology, Kanazawa University, Kanazawa 9201192, JapanOptical flow estimation is a fundamental and long-standing task in computer vision, facilitating the understanding of motion within visual scenes. In this study, we aim to improve optical flow estimation, particularly in challenging scenarios involving small and fast-moving objects. Specifically, we proposed a learning-based model incorporating two key components: the Hierarchical Motion Field Alignment module, which ensures accurate estimation of objects of varying sizes while maintaining manageable computational complexity, and the Correlation Self-Attention module, which effectively handles large displacements, making the model suitable for scenarios with fast-moving objects. Additionally, we introduced a Multi-Scale Correlation Search layer to enhance the four-dimensional cost volume, enabling the model to address various types of motion. Experimental results demonstrate that our model achieves superior generalization performance and significantly improves the estimation of small, fast-moving objects.https://www.mdpi.com/1424-8220/25/9/2653attention mechanismscomputer visioncorrelationdeep learningimage processingmotion estimation
spellingShingle Dianbo Ma
Kousuke Imamura
Ziyan Gao
Xiangjie Wang
Satoshi Yamane
Hierarchical Motion Field Alignment for Robust Optical Flow Estimation
Sensors
attention mechanisms
computer vision
correlation
deep learning
image processing
motion estimation
title Hierarchical Motion Field Alignment for Robust Optical Flow Estimation
title_full Hierarchical Motion Field Alignment for Robust Optical Flow Estimation
title_fullStr Hierarchical Motion Field Alignment for Robust Optical Flow Estimation
title_full_unstemmed Hierarchical Motion Field Alignment for Robust Optical Flow Estimation
title_short Hierarchical Motion Field Alignment for Robust Optical Flow Estimation
title_sort hierarchical motion field alignment for robust optical flow estimation
topic attention mechanisms
computer vision
correlation
deep learning
image processing
motion estimation
url https://www.mdpi.com/1424-8220/25/9/2653
work_keys_str_mv AT dianboma hierarchicalmotionfieldalignmentforrobustopticalflowestimation
AT kousukeimamura hierarchicalmotionfieldalignmentforrobustopticalflowestimation
AT ziyangao hierarchicalmotionfieldalignmentforrobustopticalflowestimation
AT xiangjiewang hierarchicalmotionfieldalignmentforrobustopticalflowestimation
AT satoshiyamane hierarchicalmotionfieldalignmentforrobustopticalflowestimation