跳到主要导航 跳到搜索 跳到主要内容

Adaptive Registration and Reference-Driven Spatial Attention Network for Super-Resolution of Multi-frame Infrared Images

  • Junyao Zhao
  • , Can Cui
  • , Jun Ke*
  • *此作品的通讯作者
  • Beijing Institute of Technology
  • Tsinghua University

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

摘要

Due to hardware limitations infrared images typically suffer from low resolution and low signal-to-noise ratio. Multi-frame image super-resolution technology exploits the additional information from micro-displacements in multiple low-resolution images to improve image resolution. In this work, we have designed an adaptive registration and reference-driven spatial attention network(ARRSAN) for multi-frame super-resolution. Additionally, a denoising module is integrated into the registration process to mitigate the impact of read shot noise on the registration and super-resolution processes. Results from three datasets, FLIR, Rafael and SIATD demonstrate the performance of our method, even in the presence of noise.

源语言英语
主期刊名Holography, Diffractive Optics, and Applications XIV
编辑Changhe Zhou, Ting-Chung Poon, Liangcai Cao, Hiroshi Yoshikawa
出版商SPIE
ISBN(电子版)9781510682085
DOI
出版状态已出版 - 2024
活动Holography, Diffractive Optics, and Applications XIV 2024 - Nantong, 中国
期限: 12 10月 2024 → 15 10月 2024

丛书

姓名Proceedings of SPIE - The International Society for Optical Engineering
卷13240
ISSN(印刷版)0277-786X
ISSN(电子版)1996-756X

会议

会议Holography, Diffractive Optics, and Applications XIV 2024
国家/地区中国
市Nantong
时期12/10/24 → 15/10/24

学术指纹

探究 'Adaptive Registration and Reference-Driven Spatial Attention Network for Super-Resolution of Multi-frame Infrared Images' 的科研主题。它们共同构成独一无二的学术指纹。

引用此