Image quality of compressive single-pixel imaging using different Hadamard orderings

Vaz, Pedro G.; Amaral, Daniela Patti do; Ferreira, L.; Morgado, Miguel; Cardoso, J. M. R.

doi:10.1364/oe.387612

Cited by 100 publications

(40 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Compared with the random patterns, different deterministic patterns (e.g., Hadamard 33 – 35 and Fourier 36 patterns) have been proposed and demonstrated their advantages in computational GI configuration with a low sampling ratio due to their characteristic (e.g., orthogonality). Nevertheless, the advantages of deterministic patterns are always accompanied by some limitations (e.g., the inapplicability in passive illumination cases, the size limitation of Hadamard matrix and the order effect of Hadamard patterns 34 , 35 ), which complicates the pattern effect on GI efficiency. For simplicity and without loss of generality, we here choose random patterns to focus on the study of GI algorithm effect.…”

Section: Resultsmentioning

confidence: 99%

Imaging reconstruction comparison of different ghost imaging algorithms

Liu

2020

Sci Rep

View full text Add to dashboard Cite

As an indirect and computational imaging approach, imaging reconstruction efficiency is critical for ghost imaging (GI). Here, we compare different GI algorithms, including logarithmic GI and exponential GI we proposed, by numerically analysing their imaging reconstruction efficiency and error tolerance. Simulation results show that compressive GI algorithm has the highest reconstruction efficiency due to its global optimization property. Error tolerance studies further manifest that compressive GI and exponential GI are sensitive to the error ratio. By replacing the bucket input of compressive GI with different bucket object signal functions, we integrate compressive GI with other GI algorithms and discuss their imaging efficiency. With the combination between the differential GI (or normalized GI) and compressive GI, both reconstruction efficiency and error tolerance will present the best performance. Moreover, an optical encryption is proposed by combining logarithmic GI, exponential GI and compressive GI, which can enhance the encryption security based on GI principle.

show abstract

Section: Resultsmentioning

confidence: 99%

Imaging reconstruction comparison of different ghost imaging algorithms

Liu

2020

Sci Rep

View full text Add to dashboard Cite

show abstract

“…To be consistent with the general significance of natural scenes, several orderings of Hadamard bases have been demonstrated previously. Depending on the uniqueness of each sample, these orderings exhibit slightly different performance [45,50]. In this work, we chose to use a square sampling path (detailed in the supplement).…”

Section: Methodsmentioning

confidence: 99%

Imaging biological tissue with high-throughput single-pixel compressive holography

Wu¹,

Luo

Huang

et al. 2021

Preprint

View full text Add to dashboard Cite

Single-pixel holography (SPH) is capable of generating holographic images with rich spatial information by employing only a single-pixel detector. Thanks to the relatively low dark-noise production, high sensitivity, large bandwidth, and cheap price of single-pixel detectors in comparison to pixel-array detectors, SPH is becoming an attractive imaging modality at wavelengths where pixel-array detectors are not available or prohibitively expensive. Moreover, SPH is particularly advantageous when imaging through scattering media or in scarce illumination with compressive sensing. In the current practice of SPH, the throughput of the system is mainly limited by the phase-encoded illumination and the ways to realize phase stepping. In this work, we developed a high-through single-pixel compressive holography, achieving a space-bandwidth-time product (SBP-T) of 41,667 pixels/s. This result indicates that by using a single-pixel detector, information of holographic images containing up to 65,536 pixels can be collected within only 3 seconds. The high-throughput was realized by enabling phase stepping naturally in time and abandoning the need for phase-encoded illumination. We further show that compressive sensing can be conveniently adapted to significantly reduce the acquisition time. Besides being high throughput, we also show that this holographic system is scalable to provide either a large field of view (~83 mm2) or a high resolution (5.8 μm × 4.3 μm). In particular, high-resolution holographic images of a piece of rat tail were presented, exhibiting rich information of mussel, cortical bone, and cancellous bone. Given that microscopic images of biological tissue has rarely been explored in the current practice of SPH, we anticipate the developed high-throughput SPH is promising to nourish the development of multi-spectrum imaging by providing high-quality holographic images for biological tissues.

show abstract

“…Regarding gesture recognition, it is enough to keep the resolution low. To reduce the number of illuminations, illumination methods [11][12][13], calculation methods [14,15] and their combination [16,17] have been proposed. Then, the use of deep learning for restoration of the original image from data reconstructed with a small number of illuminations has been largely explored in the last five years.…”

Section: Introductionmentioning

confidence: 99%

Deep-Learning-Assisted Single-Pixel Imaging for Gesture Recognition in Consideration of Privacy

Mukojima

Yasugi

Mizutani

et al. 2022

IEICE Trans. Electron.

View full text Add to dashboard Cite

We have utilized single-pixel imaging and deep-learning to solve the privacy-preserving problem in gesture recognition for interactive display. Silhouette images of hand gestures were acquired by use of a display panel as an illumination. Reconstructions of gesture images have been performed by numerical experiments on single-pixel imaging by changing the number of illumination mask patterns. For the training and the image restoration with deep learning, we prepared reconstructed data with 250 and 500 illuminations as datasets. For each of the 250 and 500 illuminations, we prepared 9000 datasets in which original images and reconstructed data were paired. Of these data, 8500 data were used for training a neural network (6800 data for training and 1700 data for validation), and 500 data were used to evaluate the accuracy of image restoration. Our neural network, based on U-net, was able to restore images close to the original images even from reconstructed data with greatly reduced number of illuminations, which is 1/40 of the single-pixel imaging without deep learning. Compared restoration accuracy between cases using shadowgraph (black on white background) and negative-positive reversed images (white on black background) as silhouette image, the accuracy of the restored image was lower for negative-positive-reversed images when the number of illuminations was small. Moreover, we found that the restoration accuracy decreased in the order of rock, scissor, and paper. Shadowgraph is suitable for gesture silhouette, and it is necessary to prepare training data and construct neural networks, to avoid the restoration accuracy between gestures when further reducing the number of illuminations.

show abstract

Image quality of compressive single-pixel imaging using different Hadamard orderings

Cited by 100 publications

References 34 publications

Imaging reconstruction comparison of different ghost imaging algorithms

Imaging reconstruction comparison of different ghost imaging algorithms

Imaging biological tissue with high-throughput single-pixel compressive holography

Deep-Learning-Assisted Single-Pixel Imaging for Gesture Recognition in Consideration of Privacy

Contact Info

Product

Resources

About