Image reconstruction through dynamic scattering media based on deep learning

Sun, Yiwei; Shi, Jianhong; Sun, Lei; Fan, Jianping; Zeng, Guihua

doi:10.1364/oe.27.016032

Cited by 112 publications

(30 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Image deblurring can be considered as a special image-to-image conversion (Yeh et al, 2016). In Sun et al (2019), Ramakrishnan et al (2017), Kupyn et al (2018), Arjovsky et al (2017), and Johnson et al (2016), image deblurring based on generative adversarial networks optimized by different methods was proposed. In Ramakrishnan et al (2017), a novel deep filter based on GAN integrated with global skip connection and dense architecture is proposed to remove blur caused by the relative motion between the camera and the object in 3D space.…”

Section: Related Workmentioning

confidence: 99%

RFID multi-tag dynamic detection measurement based on conditional generative adversarial nets and YOLOv3

Lin

Liu

et al. 2021

Transactions of the Institute of Measurement and Control

View full text Add to dashboard Cite

The quality of multi-tag imaging greatly affects the effective detection of multi-tag. When multi-tag moves rapidly, the image may have serious dynamic blur, and tags can not be detected efficiently. In previous work, it is generally assumed that blur kernel and noise stationary to improve image quality. However, the dynamic deblurring of Radio Frequency Identification (RFID) multi-tag imaging is an ill-posed inverse problem. In this paper, firstly, blur-sharp multi-tag image pairs are made by superimposing and averaging the adjoin random frames. Then, we propose blind deblurring for dynamic RFID multi-tag imaging based on conditional generative adversarial nets (CGANs), which adds perceptual loss and content loss to generator to make image sharper. Finally, tags are detected by YOLOv3 in real time in end-to-end manner. Experimental results demonstrate that PSNR is at least 0.56dB higher and speed is at least 31.25 % faster than that of the current improved convolution neural networks (CNN). CGANs can remove image blur better, which has great superiority in the field of dynamic multi-tag imaging. In addition, YOLOv3 detects multi-tag quickly, thereby improving the detection accuracy.

show abstract

Section: Related Workmentioning

confidence: 99%

RFID multi-tag dynamic detection measurement based on conditional generative adversarial nets and YOLOv3

Lin

Liu

et al. 2021

Transactions of the Institute of Measurement and Control

View full text Add to dashboard Cite

show abstract

“…The dataset 1-5 are named respectively as the no-overlap complex object dataset, the overlap complex object dataset, the media of same property dataset, the media of different property dataset, the human face dataset. The cardinal targets commonly used in speckle correlation imaging based on deep learning are handwritten characters [10][11][12][13][14][15][16] . In order to improve the complexity of the targets to be restored in the first four datasets, the handwritten characters in MNIST are randomly combined in pairs.…”

Section: Simulationmentioning

confidence: 99%

Learning-based method to reconstruct complex targets through scattering medium beyond the memory effect

Guo¹,

Zhu

Sun³

et al. 2020

Opt. Express

View full text Add to dashboard Cite

Strong scattering medium brings great difficulties to optical imaging, which is also a problem in medical imaging and many other fields. Optical memory effect makes it possible to image through strong random scattering medium. However, this method also has the limitation of limited angle field-of-view (FOV), which prevents it from being applied in practice. In this paper, a kind of practical convolutional neural network called PDSNet is proposed, which effectively breaks through the limitation of optical memory effect on FOV. Experiments is conducted to prove that the scattered pattern can be reconstructed accurately in real-time by PDSNet, and it is widely applicable to retrieve complex objects of random scales and different scattering media.

show abstract

“…OCT and wavefront shaping techniques demand a sophisticated optical design and hardware, which is hard to deploy in practice. In recent years, deep learning (DL) has shown favorable achievements in the field of imaging through scattering media [19][20][21][22][23][24][25][26] . Li built a "one to all" model based on the UNet backbone, which can learn statistical information about similar scattering media with different microstructures and extract statistical invariance of the speckle 23 .…”

Section: Introductionmentioning

confidence: 99%

Intensity and phase imaging through scattering media via deep despeckle complex neural networks

Liu

Sha

et al. 2022

Preprint

View full text Add to dashboard Cite

The existence of a scattering medium causes the degeneration of intensity and phase information, especially in biological imaging. The present techniques to address this challenge only focus on the reconstruction of intensity information, yet few attempts have tried to recover the phase information. We propose a method to simultaneously predict both intensity and phase information from a speckle image employing a deep despeckle complex neural network (DespeckleNet). Our method enables the high contrast single-shot imaging of biological samples through scattering media without labeling. Various experiments demonstrate the superior reconstruction and generalization performance of our method under multiple types of biological samples with different scattering media. We also provide the real-time observation of living cellular activities without any contaminations or damages to the cells. Our method offers simple yet effective imaging through scattering media and paves the way for real-time unlabeled biological imaging.

show abstract

Image reconstruction through dynamic scattering media based on deep learning

Cited by 112 publications

References 19 publications

RFID multi-tag dynamic detection measurement based on conditional generative adversarial nets and YOLOv3

RFID multi-tag dynamic detection measurement based on conditional generative adversarial nets and YOLOv3

Learning-based method to reconstruct complex targets through scattering medium beyond the memory effect

Intensity and phase imaging through scattering media via deep despeckle complex neural networks

Contact Info

Product

Resources

About