Foveated ghost imaging based on deep learning

Zhai, Xiang; Cheng, Zhengdong; Hu, Yangdi; Chen, Yi; Liang, Zhenyu; Yuan, Wei

doi:10.1016/j.optcom.2019.05.019

Cited by 26 publications

(4 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Step 3: According to equation (11), the compressed JPS of single image is translated by joint power spectral division multiplexing, which realizes the superposition of multiple images without crosstalk. The result is expressed as J (u, v), which is secondary ciphertext.…”

Section: Encryption Processmentioning

confidence: 99%

“…Jiao et al proposed two novel visual cryptography (VC) schemes by combining VC with SPI, which improved the security of the encryption system [10]. Zhai et al combined foveated speckle pattern with GAN-based GI object detection system to realize selecting the region of interest for foveated imaging intelligently, which improved the reconstruction accuracy [11].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Research on multiple-image encryption scheme based on joint power spectral division multiplexing and ghost imaging

Wang¹,

Zhang²,

Zhang³

et al. 2021

Laser Phys.

View full text Add to dashboard Cite

Since there are problems of easy cross-talk, large ciphertext transmission and low security in the process of multiple-image encryption, in order to solve these problems, a multiple-image encryption algorithm based on joint power spectral division multiplexing and ghost imaging (GI) is proposed. The joint transform correlator is combined with GI to realize ‘one encryption to one key’, which improves the security of the encryption system. Joint power spectrum (JPS) is compressed by the iterative restoration algorithm to reduce the transmissions of ciphertext. The joint power spectral division multiplexing is used, and the optimized phase mask is linearly superposed. The JPS of each channel has different positions on the spectrum plane, and then the non-crosstalk superposition is realized by window filtering. In this paper, the security, robustness, and encryption capacity of the encryption system are verified by numerical simulation.

show abstract

Section: Encryption Processmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Research on multiple-image encryption scheme based on joint power spectral division multiplexing and ghost imaging

Wang¹,

Zhang²,

Zhang³

et al. 2021

Laser Phys.

View full text Add to dashboard Cite

show abstract

“…Sui et al proposed a multi-image authentication method by combining the intensity equation transmission technique [14]; Zhang et al combined phase recovery with correlation imaging and proposed a multi-image based on phase recovery algorithm and ghost imaging holographic encryption technique [15]; Yuan et al proposed a multi-image encryption scheme based on Hadamard matrix ghost imaging and spatial multiplexing, which uses spatial multiplexing technique to combine multiple sampled images into one image, and then performs Fourier transform inverse operation, and finally, the combined image is encrypted with ghost images [16]. Currently, these multi-image encryption methods increase the number of encrypted images but also increase the complexity of the system; at the same time, the time and complexity of data processing increases with the increase of encryption capacity [15,[17][18][19][20].…”

Section: Introductionmentioning

confidence: 99%

Research on multi-image encryption method based on image scaling and ghost imaging

Qiu,

Zhang,

Zhang

2024

Laser Phys.

View full text Add to dashboard Cite

Optical information processing technology itself is characterized by high speed and parallelism, while the wavelength of light is short and the loading information capacity is large. Therefore, optical encryption is of great significance in the field of image encryption. For multi-image encryption, a multi-image encryption system (ISK-MGI) based on image scaling and ghost imaging is proposed in this paper. In the encryption process, image scaling is used to embed the image information to be encrypted into the artifact image to achieve the first step of artifact encryption; then the ghost imaging encryption is used to encrypt any of the artifact images and the intensity sequence obtained from the barrel detector is used as the cipher text; the integration property of Fourier transform is used to obtain the modulation patterns of other artifact images as the key. The feasibility, security and robustness of the scheme are verified by experiments and simulations. The scheme not only solves the image information crosstalk problem of multi-image encryption, but also has high security. Each image in this method has a primary public key and a secondary private key, which solves the potential risk of sharing public keys and provides a new idea for optical information encryption.

show abstract

“…Recently, the deep learning (DL) technique is employed to identify images [26,27] and improve the quality of images with the deep neural network (DNN) [28][29][30][31][32][33][34][35][36]. Specifically, computational ghost imaging via deep learning (CGIDL) has shown a minimum ratio of Nyquist limit down to ∼ 5% [29,33].…”

Section: Introductionmentioning

confidence: 99%

0.8% Nyquist computational ghost imaging via non-experimental deep learning

Song,

Nie,

et al. 2021

Preprint

View full text Add to dashboard Cite

We present a framework for computational ghost imaging based on deep learning and customized pink noise speckle patterns. The deep neural network in this work, which can learn the sensing model and enhance image reconstruction quality, is trained merely by simulation. To demonstrate the sub-Nyquist level in our work, the conventional computational ghost imaging results, reconstructed imaging results using white noise and pink noise via deep learning are compared under multiple sampling rates at different noise conditions. We show that the proposed scheme can provide highquality images with a sampling rate of 0.8% even when the object is outside the training dataset, and it is robust to noisy environments. This method is excellent for various applications, particularly those that require a low sampling rate, fast reconstruction efficiency, or experience strong noise interference.

show abstract

Foveated ghost imaging based on deep learning

Cited by 26 publications

References 24 publications

Research on multiple-image encryption scheme based on joint power spectral division multiplexing and ghost imaging

Research on multiple-image encryption scheme based on joint power spectral division multiplexing and ghost imaging

Research on multi-image encryption method based on image scaling and ghost imaging

0.8% Nyquist computational ghost imaging via non-experimental deep learning

Contact Info

Product

Resources

About