基于EMNIST训练的低采样计算重影成像高效神经网络

IF 1.2 4区物理与天体物理 Q4 OPTICS

Journal of Modern Optics Pub Date : 2022-11-01 DOI:10.1080/09500340.2022.2138594

Xu Chen, Chunfang Wang, Quanchao Zhao

{"title":"基于EMNIST训练的低采样计算重影成像高效神经网络","authors":"Xu Chen, Chunfang Wang, Quanchao Zhao","doi":"10.1080/09500340.2022.2138594","DOIUrl":null,"url":null,"abstract":"In this paper, we propose a new network to achieve low-sampling computational ghost imaging. The proposed neural network that combines ResNeXt, eHoloNet and spatial attention mechanism is efficient for object reconstruction just based on EMNIST training, which is much simpler and time-saving. The generalization of the neural network is verified with multi-slits as well as complex object. Both simulated and experimental results show that the proposed network can give an effective reconstruction at 0.7% sampling rate. The neural network in this work is of great significance to ghost imaging in a wider application scenarios, such as real-time imaging and dynamic detection of motion object.","PeriodicalId":16426,"journal":{"name":"Journal of Modern Optics","volume":"69 1","pages":"1079 - 1085"},"PeriodicalIF":1.2000,"publicationDate":"2022-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":"{\"title\":\"An efficient neural network for low sampling computational ghost imaging based on EMNIST training\",\"authors\":\"Xu Chen, Chunfang Wang, Quanchao Zhao\",\"doi\":\"10.1080/09500340.2022.2138594\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"In this paper, we propose a new network to achieve low-sampling computational ghost imaging. The proposed neural network that combines ResNeXt, eHoloNet and spatial attention mechanism is efficient for object reconstruction just based on EMNIST training, which is much simpler and time-saving. The generalization of the neural network is verified with multi-slits as well as complex object. Both simulated and experimental results show that the proposed network can give an effective reconstruction at 0.7% sampling rate. The neural network in this work is of great significance to ghost imaging in a wider application scenarios, such as real-time imaging and dynamic detection of motion object.\",\"PeriodicalId\":16426,\"journal\":{\"name\":\"Journal of Modern Optics\",\"volume\":\"69 1\",\"pages\":\"1079 - 1085\"},\"PeriodicalIF\":1.2000,\"publicationDate\":\"2022-11-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"2\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Modern Optics\",\"FirstCategoryId\":\"101\",\"ListUrlMain\":\"https://doi.org/10.1080/09500340.2022.2138594\",\"RegionNum\":4,\"RegionCategory\":\"物理与天体物理\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"OPTICS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Modern Optics","FirstCategoryId":"101","ListUrlMain":"https://doi.org/10.1080/09500340.2022.2138594","RegionNum":4,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"OPTICS","Score":null,"Total":0}

引用次数: 2

摘要

在本文中，我们提出了一种新的网络来实现低采样计算重影成像。所提出的将ResNeXt、eHoloNet和空间注意力机制相结合的神经网络仅基于EMNIST训练就可以有效地进行对象重建，而且更简单、省时。通过多狭缝和复杂对象验证了神经网络的泛化能力。仿真和实验结果表明，该网络在0.7%的采样率下可以进行有效的重构。这项工作中的神经网络对重影成像在更广泛的应用场景中具有重要意义，如运动物体的实时成像和动态检测。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

An efficient neural network for low sampling computational ghost imaging based on EMNIST training

In this paper, we propose a new network to achieve low-sampling computational ghost imaging. The proposed neural network that combines ResNeXt, eHoloNet and spatial attention mechanism is efficient for object reconstruction just based on EMNIST training, which is much simpler and time-saving. The generalization of the neural network is verified with multi-slits as well as complex object. Both simulated and experimental results show that the proposed network can give an effective reconstruction at 0.7% sampling rate. The neural network in this work is of great significance to ghost imaging in a wider application scenarios, such as real-time imaging and dynamic detection of motion object.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

Journal of Modern Optics 物理-光学

CiteScore

2.90

自引率

0.00%

发文量

审稿时长

2.6 months

期刊介绍： The journal (under its former title Optica Acta) was founded in 1953 - some years before the advent of the laser - as an international journal of optics. Since then optical research has changed greatly; fresh areas of inquiry have been explored, different techniques have been employed and the range of application has greatly increased. The journal has continued to reflect these advances as part of its steadily widening scope. Journal of Modern Optics aims to publish original and timely contributions to optical knowledge from educational institutions, government establishments and industrial R&D groups world-wide. The whole field of classical and quantum optics is covered. Papers may deal with the applications of fundamentals of modern optics, considering both experimental and theoretical aspects of contemporary research. In addition to regular papers, there are topical and tutorial reviews, and special issues on highlighted areas. All manuscript submissions are subject to initial appraisal by the Editor, and, if found suitable for further consideration, to peer review by independent, anonymous expert referees. General topics covered include: • Optical and photonic materials (inc. metamaterials) • Plasmonics and nanophotonics • Quantum optics (inc. quantum information) • Optical instrumentation and technology (inc. detectors, metrology, sensors, lasers) • Coherence, propagation, polarization and manipulation (classical optics) • Scattering and holography (diffractive optics) • Optical fibres and optical communications (inc. integrated optics, amplifiers) • Vision science and applications • Medical and biomedical optics • Nonlinear and ultrafast optics (inc. harmonic generation, multiphoton spectroscopy) • Imaging and Image processing