Route to Intelligent Imaging Reconstruction via Terahertz Nonlinear Ghost Imaging

Terahertz Sensing of L-Valine and L-Phenylalanine Solutions

To detect and differentiate two essential amino acids (L-Valine and L-Phenylalanine) in the human body, a novel asymmetrically folded dual-aperture metal ring terahertz metasurface sensor was designed. A solvent mixture of water and glycerol with a volume ratio of 2:8 was proposed to reduce the absorption of terahertz waves by reducing the water content. A sample chamber with a controlled liquid thickness of 15 μm was fabricated. And a terahertz time-domain spectroscopy (THz-TDS) system, which is capable of horizontally positioning the samples, was assembled. The results of the sensing test revealed that as the concentration of valine solution varied from 0 to 20 mmol/L, the sensing resonance peak shifted from 1.39 THz to 1.58 THz with a concentration sensitivity of 9.98 GHz/mmol∗L-1. The resonance peak shift phenomenon in phenylalanine solution was less apparent. It is assumed that the coupling enhancement between the absorption peak position of solutes in the solution and the sensing peak position amplified the terahertz localized electric field resonance, which resulted in the increase in frequency shift. Therefore, it could be shown that the sensor has capabilities in performing the marker sensing detection of L-Valine.

Fast high quality computational ghost imaging based on saliency variable sampling detection

Fast computational ghost imaging with high quality and ultra-high-definition resolution reconstructed images has important application potential in target tracking, biological imaging and other fields. However, as far as we know, the resolution (pixels) of the reconstructed image is related to the number of measurements. And the limited resolution of reconstructed images at low measurement times hinders the application of computational ghost imaging. Therefore, in this work, a new computational ghost imaging method based on saliency variable sampling detection is proposed to achieve high-quality imaging at low measurement times. This method physically variable samples the salient features and realizes compressed detection of computational ghost imaging based on the salient periodic features of the bucket detection signal. Numerical simulation and experimental results show that the reconstructed image quality of our method is similar to the compressed sensing method at low measurement times. Even at 500 (sampling rate 0.76 % ) measurement times, the reconstructed image of the method still has the target features. Moreover, the 2160 × 4096 (4K) pixels ultra-high-definition resolution reconstructed images can be obtained at only a sampling rate of 0.11 % . This method has great potential value in real-time detection and tracking, biological imaging and other fields.

Adaptive locating foveated ghost imaging based on affine transformation

Ghost imaging (GI) has been widely used in the applications including spectral imaging, 3D imaging, and other fields due to its advantages of broad spectrum and anti-interference. Nevertheless, the restricted sampling efficiency of ghost imaging has impeded its extensive application. In this work, we propose a novel foveated pattern affine transformer method based on deep learning for efficient GI. This method enables adaptive selection of the region of interest (ROI) by combining the proposed retina affine transformer (RAT) network with minimal computational and parametric quantities with the foveated speckle pattern. For single-target and multi-target scenarios, we propose RAT and RNN-RAT (recurrent neural network), respectively. The RAT network enables an adaptive alteration of the fovea of the variable foveated patterns spot to different sizes and positions of the target by predicting the affine matrix with a minor number of parameters for efficient GI. In addition, we integrate a recurrent neural network into the proposed RAT to form an RNN-RAT model, which is capable of performing multi-target ROI detection. Simulations and experimental results show that the method can achieve ROI localization and pattern generation in 0.358 ms, which is a 1 × 105 efficiency improvement compared with the previous methods and improving the image quality of ROI by more than 4 dB. This approach not only improves its overall applicability but also enhances the reconstruction quality of ROI. This creates additional opportunities for real-time GI.

Terahertz Spatiotemporal Wave Synthesis in Random Systems

Complex media have emerged as a powerful and robust framework to control light-matter interactions designed for task-specific optical functionalities. Studies on wavefront shaping through disordered systems have demonstrated optical wave manipulation capabilities beyond conventional optics, including aberration-free and subwavelength focusing. However, achieving arbitrary and simultaneous control over the spatial and temporal features of light remains challenging. In particular, no practical solution exists for field-level arbitrary spatiotemporal control of wave packets. A new paradigm shift has emerged in the terahertz frequency domain, offering methods for absolute time-domain measurements of the scattered electric field, enabling direct field-based wave synthesis. In this work, we report the experimental demonstration of field-level control of single-cycle terahertz pulses on arbitrary spatial points through complex disordered media.

Lensless single-fiber ghost imaging

We demonstrate lensless single-fiber ghost imaging, which allows illumination and collection using a single optical fiber without a transmission-type system. Speckle patterns with relative coincidence degrees of 0.14 were formed by image reconstruction using improved differential ghost imaging. Employing fiber with a diameter of 105 µm, we achieved a spatial resolution of 0.05 mm in an observing area of 9m m 2, at a working distance of 10 mm. Compared to a conventional neuroendoscope at a power density of 94m W/c m 2, our imaging could be realized by extremely weak illumination at a laser power density of 0.10m W/c m 2. Using our lensless single-fiber ghost imaging, with 30,000 speckle patterns and implementing a diffuser, we attained an average coincidence degree of 0.45.

Research on data-driven low-sampling-rate single-pixel imaging method

Single-pixel imaging requires only a unit detector with no spatial resolution capability to acquire spatial information of the target and reconstruct the image. However, the quality of reconstructing images strongly depends on measurement matrices and their number of samples, making it challenging to achieve high-quality imaging with fewer samples. In this Letter, a dataset-driven low-sampling-rate single-pixel imaging method is proposed. It utilizes a network model driven by the image datasets to directly extract target feature information from a small number of samples and reconstruct the image. Experimental results demonstrate that, compared to traditional single-pixel imaging methods, this method no longer depends strongly on the relationship between the measurement matrices and the samples, and it can achieve an ideal imaging effect with a structural similarity of 90.20% at low sampling rates.

High-throughput terahertz imaging: progress and challenges

Many exciting terahertz imaging applications, such as non-destructive evaluation, biomedical diagnosis, and security screening, have been historically limited in practical usage due to the raster-scanning requirement of imaging systems, which impose very low imaging speeds. However, recent advancements in terahertz imaging systems have greatly increased the imaging throughput and brought the promising potential of terahertz radiation from research laboratories closer to real-world applications. Here, we review the development of terahertz imaging technologies from both hardware and computational imaging perspectives. We introduce and compare different types of hardware enabling frequency-domain and time-domain imaging using various thermal, photon, and field image sensor arrays. We discuss how different imaging hardware and computational imaging algorithms provide opportunities for capturing time-of-flight, spectroscopic, phase, and intensity image data at high throughputs. Furthermore, the new prospects and challenges for the development of future high-throughput terahertz imaging systems are briefly introduced.

High-fidelity and high-robustness free-space ghost transmission in complex media with coherent light source using physics-driven untrained neural network

It is well recognized that it is challenging to realize high-fidelity and high-robustness ghost transmission through complex media in free space using coherent light source. In this paper, we report a new method to realize high-fidelity and high-robustness ghost transmission through complex media by generating random amplitude-only patterns as 2D information carriers using physics-driven untrained neural network (UNN). The random patterns are generated to encode analog signals (i.e., ghost) without any training datasets and labeled data, and are used as information carriers in a free-space optical channel. Coherent light source modulated by the random patterns propagates through complex media, and a single-pixel detector is utilized to collect light intensities at the receiving end. A series of optical experiments have been conducted to verify the proposed approach. Experimental results demonstrate that the proposed method can realize high-fidelity and high-robustness analog-signal (ghost) transmission in complex environments, e.g., around a corner, or dynamic and turbid water. The proposed approach using the designed physics-driven UNN could open an avenue for high-fidelity free-space ghost transmission through complex media.

Holographic THz Beam Generation by Nonlinear Plasmonic Metasurface Emitters

The advancement of terahertz (THz) technology hinges on the progress made in the development of efficient sources capable of generating and shaping the THz emission. However, the currently available THz sources provide limited control over the generated field. Here, we use near-field interactions in nonlinear Pancharatnam-Berry phase plasmonic metasurfaces to achieve deep subwavelength, precise, and continuous control over the local amplitude of the emitted field. We show that this new ability can be used for holographic THz beam generation. Specifically, we demonstrate the generation of precisely shaped Hermite-Gauss, Top-Hat, and triangular beams. We show that using this method, higher-order modes are completely suppressed, indicating optimal nonlinear diffraction efficiency. In addition, we demonstrate the application of the generated structured beams for obtaining enhanced imaging resolution and contrast. These demonstrations hold immense potential to address challenges associated with a broad range of new applications employing THz technology.

Elimination of signal amplitude disturbance in ghost imaging using an auxiliary laser channel

Ghost imaging can be used to detect objects in a nonstationary environment or in the presence of variable ambient light, making it attractive when conventional imaging methods are ineffective. However, the conventional ghost imaging algorithm is susceptible to temporal fluctuations in the detected signal. In this work, we propose a polarization-multiplexed auxiliary laser channel propagating along the same optical path with the main one. The signal in the auxiliary channel is used as a reference and allows the elimination of signal disturbance. A quantitative analysis and comparison of the proposed method's performance to the high-pass filtering method are demonstrated. For an illumination pattern refresh rate of 10 Hz, effective suppression of bucket signal fluctuations has been experimentally demonstrated. For a disturbance frequency from 1 Hz to 10 Hz, the auxiliary channel method demonstrated a ghost image Pearson correlation coefficient (PCC) of not less than 0.70, while the high-pass filtering method showed a PCC sharp drop from 0.65 to 0.02.

Spatial and spectral characteristics in realizations of broadband terahertz spectroscopy on a subwavelength scale

In this work, we take aim at the fundamental challenge for realizations of broadband terahertz (THz) spectroscopy on a subwavelength scale. We introduce apertured THz microjets in this effort to resolve the fundamental limits of spatial resolution and spectral bandwidth. The THz microjets are formed as intense foci at the rear of engineered (microcomposite) spheres and are coupled through subwavelength (circular) apertures. Such coupling enables effective transmission of THz power through samples with broad spectral bandwidths and fine spatial resolutions. We show that the apertures function as high-pass filters, with their diameter d enabling strong transmission above a cutoff frequency f_c. Our theoretical and experimental results reveal that the values for d and f_c are prescribed by a fixed spatial-spectral product df_c, whereby reductions in d (to improve the spatial resolution) can raise f_c into the targeted spectrum (at the expense of spectral bandwidth). We use this understanding to demonstrate broadband (0.3-0.7 THz) THz spectroscopy of lactose at the subwavelength (365 µm) scale. These results for apertured THz microjets represent a 20-fold improvement in spatial resolution over analogous apertured THz plane waves. Overall, our findings show promise for studies of carcinogenesis, pathogenesis, and the like.

High-resolution self-corrected single-pixel imaging through dynamic and complex scattering media

Imaging with single-pixel detectors becomes attractive in many applications where pixelated detectors are not available or cannot work. Based on a correlation between the probing patterns and the realizations, optical imaging with single-pixel detector offers an indirect way to recover a sample. It is well recognized that single-pixel optical imaging through dynamic and complex scattering media is challenging, and dynamic scaling factors lead to serious mismatches between the probing patterns and the realizations. In this paper, we report self-corrected imaging to realize high-resolution object reconstruction through dynamic and complex scattering media using a parallel detection with dual single-pixel detectors. The proposed method can supervise and self-correct dynamic scaling factors, and can implement high-resolution object reconstruction through dynamic and complex scattering media where conventional methods could not work. Spatial resolution of 44.19 µm is achieved which approaches diffraction limit (40.0 µm) in the designed optical setup. The achievable spatial resolution is dependent on pixel size of spatial light modulator. It is experimentally validated that the proposed method shows unprecedented robustness against complex scattering. The proposed self-corrected imaging provides a solution for ghost recovery, enabling high-resolution object reconstruction in complex scattering environments.

Hanbury Brown-Twiss approach for imaging through a dynamic scattering medium

The Hanbury Brown-Twiss approach, associated with the correlation of intensity fluctuations at two different points in a wave field, unveils fundamental aspects of light. Here, we propose and experimentally demonstrate an imaging and phase recovery technique through a dynamic scattering medium using the Hanbury Brown-Twiss approach. A detailed theoretical basis is presented and verified by experimental demonstrations. To validate the application of the proposed technique, the randomness of the dynamically scattered light is exploited using temporal ergodicity for evaluating the correlation of intensity fluctuations and consequently applying it in the reconstruction of the object hidden behind the dynamic diffuser.

Terahertz Nonlinear Ghost Imaging via Plane Decomposition: Toward Near-Field Micro-Volumetry

Terahertz time-domain imaging targets the reconstruction of the full electromagnetic morphology of an object. In this spectral range, the near-field propagation strongly affects the information in the space-time domain in items with microscopic features. While this often represents a challenge, as the information needs to be disentangled to obtain high image fidelity, here, we show that such a phenomenon can enable three-dimensional microscopy. Specifically, we investigate the capability of the time-resolved nonlinear ghost imaging methodology to implement field-sensitive micro-volumetry by plane decomposition. We leverage the temporally resolved, field-sensitive detection to "refocus" an image plane at an arbitrary distance from the source, which defines the near-field condition, and within a microscopic sample. Since space-time coupling rapidly evolves and diffuses within subwavelength length scales, our technique can separate and discriminate the information originating from different planes at different depths. Our approach is particularly suitable for objects with sparse micrometric details. Building upon this principle, we demonstrate complex, time-domain volumetry resolving internal object planes with subwavelength resolution, discussing the range of applicability of our technique.

One-time readout temporal single-pixel imaging

Ghost imaging (GI) allows retrieval of the images of an uninterrogated object nonlocally based on cross correlation of photons. Central to GI is the integration of sparse detection events, i.e., bucket detection, even in the time domain. Here we report temporal single-pixel imaging of a non-integrating class as a viable GI variant that obviates the need for constant vigil. Mere division of the otherwise distorted waveforms by the known impulse response function of the detector makes the corrected waveforms readily available. This tempts us to use slow and as such less costly commercially available optoelectronic devices like light-emitting diodes and solar cells for imaging purposes on a one-time readout basis.

Feature ghost imaging for color identification

On the basis of computational ghost imaging (CGI), we present a new imaging technique, feature ghost imaging (FGI), which can convert the color information into distinguishable edge features in retrieved grayscale images. With the edge features extracted by different order operators, FGI can obtain the shape and the color information of objects simultaneously in a single-round detection using one single-pixel detector. The feature distinction of rainbow colors is presented in numerical simulations and the verification of FGI's practical performance is conducted in experiments. Furnishing a new perspective to the imaging of colored objects, our FGI extends the function and the application fields of traditional CGI while sustaining the simplicity of the experimental setup.

Indirect spectrum measurement via random phase modulation and detection in temporal domain

Spectroscopy continues to provide possibilities for a deeper understanding of fundamental physical phenomena. Traditional spectral measurement method, dispersive Fourier transformation, is always limited by its realization condition (detection in the temporal far-field). Inspired by Fourier ghost imaging, we put forward an indirect spectrum measurement to overcome the limitation. The spectrum information is reconstructed via random phase modulation and near-field detection in the time domain. Since all operations are realized in the near-field region, the required length of dispersion fiber and optical loss are greatly reduced. Considering the application in spectroscopy, the length of required dispersion fiber, the spectrum resolution, the range of spectrum measurement and the requirement on bandwidth of photodetector are investigated.

Dual-mode adaptive-SVD ghost imaging

In this paper, we present a dual-mode adaptive singular value decomposition ghost imaging (A-SVD GI), which can be easily switched between the modes of imaging and edge detection. It can adaptively localize the foreground pixels via a threshold selection method. Then only the foreground region is illuminated by the singular value decomposition (SVD) - based patterns, consequently retrieving high-quality images with fewer sampling ratios. By changing the selecting range of foreground pixels, the A-SVD GI can be switched to the mode of edge detection to directly reveal the edge of objects, without needing the original image. We investigate the performance of these two modes through both numerical simulations and experiments. We also develop a single-round scheme to halve measurement numbers in experiments, instead of separately illuminating positive and negative patterns in traditional methods. The binarized SVD patterns, generated by the spatial dithering method, are modulated by a digital micromirror device (DMD) to speed up the data acquisition. This dual-mode A-SVD GI can be applied in various applications, such as remote sensing or target recognition, and could be further extended for multi-modality functional imaging/detection.

Quantum ghost imaging based on a "looking back" 2D SPAD array

Quantum ghost imaging (QGI) is an intriguing imaging protocol that exploits photon-pair correlations stemming from spontaneous parametric down-conversion (SPDC). QGI retrieves images from two-path joint measurements, where single-path detection does not allow us to reconstruct the target image. Here we report on a QGI implementation exploiting a two-dimensional (2D) single-photon avalanche diode (SPAD) array detector for the spatially resolving path. Moreover, the employment of non-degenerate SPDC allows us to investigate samples at infrared wavelengths without the need for short-wave infrared (SWIR) cameras, while the spatial detection can be still performed in the visible region, where the more advanced silicon-based technology can be exploited. Our findings advance QGI schemes towards practical applications.

Versatile OSCAT time-domain THz spectrometer

We report on a compact and versatile time-domain spectrometer operating in the THz spectral region from 0.2 to 2.5 THz based on ultrafast Yb:CALGO laser and photo-conductive antennas. The spectrometer operates with the optical sampling by cavity tuning (OSCAT) method based on laser repetition rate tuning, which allows at the same time the implementation of a delay-time modulation scheme. The whole characterization of the instrument is presented and compared to the classical THz time-domain spectroscopy implementation. THz spectroscopic measurements on a 520-μm thick GaAs wafer substrate together with water vapor absorption measurements are also reported to further validate the instrument capabilities.

Foveated panoramic ghost imaging

Panoramic ghost imaging (PGI) is a novel method by only using a curved mirror to enlarge the field of view (FOV) of ghost imaging (GI) to 360°, making GI a breakthrough in the applications with a wide FOV. However, high-resolution PGI with high efficiency is a serious challenge because of the large amount of data. Therefore, inspired by the variant-resolution retina structure of human eye, a foveated panoramic ghost imaging (FPGI) is proposed to achieve the coexistence of a wide FOV, high resolution and high efficiency on GI by reducing the resolution redundancy, and further to promote the practical applications of GI with a wide FOV. In FPGI system, a flexible variant-resolution annular pattern structure via log-rectilinear transformation and log-polar mapping is proposed to be used for projection, which can allocate the resolution of the region of interest (ROI) and the other region of non-interest (NROI) by setting related parameters in the radial and poloidal directions independently to meet different imaging requirements. In addition, in order to reasonably reduce the resolution redundancy and avoid the loss of the necessary resolution on NROI, the variant-resolution annular pattern structure with a real fovea is further optimized to keep the ROI at any position in the center of 360° FOV by flexibly changing the initial position of the start-stop boundary on the annular pattern structure. The experimental results of the FPGI with one fovea and multiple foveae demonstrate that, compared to the traditional PGI, the proposed FPGI not only can improve the imaging quality on the ROIs with a high resolution and flexibly remain a lower-resolution imaging on the NROI with different required resolution reduction; but also reduce the reconstruction time to improve the imaging efficiency due to the reduction of the resolution redundancy.

Single-shot ultrafast terahertz photography

Multidimensional imaging of transient events has proven pivotal in unveiling many fundamental mechanisms in physics, chemistry, and biology. In particular, real-time imaging modalities with ultrahigh temporal resolutions are required for capturing ultrashort events on picosecond timescales. Despite recent approaches witnessing a dramatic boost in high-speed photography, current single-shot ultrafast imaging schemes operate only at conventional optical wavelengths, being suitable solely within an optically-transparent framework. Here, leveraging on the unique penetration capability of terahertz radiation, we demonstrate a single-shot ultrafast terahertz photography system that can capture multiple frames of a complex ultrafast scene in non-transparent media with sub-picosecond temporal resolution. By multiplexing an optical probe beam in both the time and spatial-frequency domains, we encode the terahertz-captured three-dimensional dynamics into distinct spatial-frequency regions of a superimposed optical image, which is then computationally decoded and reconstructed. Our approach opens up the investigation of non-repeatable or destructive events that occur in optically-opaque scenarios.

Low sampling high quality image reconstruction and segmentation based on array network ghost imaging

High-quality imaging under low sampling time is an important step in the practical application of computational ghost imaging (CGI). At present, the combination of CGI and deep learning has achieved ideal results. However, as far as we know, most researchers focus on one single pixel CGI based on deep learning, and the combination of array detection CGI and deep learning with higher imaging performance has not been mentioned. In this work, we propose a novel multi-task CGI detection method based on deep learning and array detector, which can directly extract target features from one-dimensional bucket detection signals at low sampling times, especially output high-quality reconstruction and image-free segmentation results at the same time. And this method can realize fast light field modulation of modulation devices such as digital micromirror device to improve the imaging efficiency by binarizing the trained floating-point spatial light field and fine-tuning the network. Meanwhile, the problem of partial information loss in the reconstructed image due to the detection unit gap in the array detector has also been solved. Simulation and experimental results show that our method can simultaneously obtain high-quality reconstructed and segmented images at sampling rate of 0.78 %. Even when the signal-to-noise ratio of the bucket signal is 15 dB, the details of the output image are still clear. This method helps to improve the applicability of CGI and can be applied to resource-constrained multi-task detection scenarios such as real-time detection, semantic segmentation, and object recognition.

Frequency-domain terahertz spectroscopy using long-carrier-lifetime photoconductive antennas

We present a telecommunication-compatible frequency-domain terahertz spectroscopy system realized by novel photoconductive antennas without using short-carrier-lifetime photoconductors. Built on a high-mobility InGaAs photoactive layer, these photoconductive antennas are designed with plasmonics-enhanced contact electrodes to achieve highly confined optical generation near the metal/semiconductor surface, which offers ultrafast photocarrier transport and, hence, efficient continuous-wave terahertz operation including both generation and detection. Consequently, using two plasmonic photoconductive antennas as a terahertz source and a terahertz detector, we successfully demonstrate frequency-domain spectroscopy with a dynamic range more than 95 dB and an operation bandwidth of 2.5 THz. Moreover, this novel approach to terahertz antenna design opens up a wide range of new possibilities for many different semiconductors and optical excitation wavelengths to be utilized, therefore bypassing short-carrier-lifetime photoconductors with limited availability.

Nonlinear field-control of terahertz waves in random media for spatiotemporal focusing

Controlling the transmission of broadband optical pulses in scattering media is a critical open challenge in photonics. To date, wavefront shaping techniques at optical frequencies have been successfully applied to control the spatial properties of multiple-scattered light. However, a fundamental restriction in achieving an equivalent degree of control over the temporal properties of a broadband pulse is the limited availability of experimental techniques to detect the coherent properties (i.e., the spectral amplitude and absolute phase) of the transmitted field. Terahertz experimental frameworks, on the contrary, enable measuring the field dynamics of broadband pulses at ultrafast (sub-cycle) time scales directly. In this work, we provide a theoretical/numerical demonstration that, within this context, complex scattering can be used to achieve spatio-temporal control of instantaneous fields and manipulate the temporal properties of single-cycle pulses by solely acting on spatial degrees of freedom of the illuminating field. As direct application scenarios, we demonstrate spatio-temporal focusing, chirp compensation, and control of the carrier-envelope-phase (CEP) of a CP-stable, transform-limited THz pulse.

Nonlinear field-control of terahertz waves in random media for spatiotemporal focusing

Controlling the transmission of broadband optical pulses in scattering media is a critical open challenge in photonics. To date, wavefront shaping techniques at optical frequencies have been successfully applied to control the spatial properties of multiple-scattered light. However, a fundamental restriction in achieving an equivalent degree of control over the temporal properties of a broadband pulse is the limited availability of experimental techniques to detect the coherent properties (i.e., the spectral amplitude and absolute phase) of the transmitted field. Terahertz experimental frameworks, on the contrary, enable measuring the field dynamics of broadband pulses at ultrafast (sub-cycle) time scales directly. In this work, we provide a theoretical/numerical demonstration that, within this context, complex scattering can be used to achieve spatio-temporal control of instantaneous fields and manipulate the temporal properties of single-cycle pulses by solely acting on spatial degrees of freedom of the illuminating field. As direct application scenarios, we demonstrate spatio-temporal focusing, chirp compensation, and control of the carrier-envelope-phase (CEP) of a CP-stable, transform-limited THz pulse.

Single-pixel three-dimensional imaging of the terahertz-wave by complex-field synthesis

We propose a novel three-dimensional (3D) imaging technique by terahertz (THz) waves. Specifically, we modulate the THz wave using diffusers to produce three different speckle-like illumination patterns. The object is raster scanned by the three illumination patterns to generate three raw images via the single-pixel detection method. Subsequently, we synthesize a complex field using the three raw images. Finally, the retrieved image is calculated using the phase correlation of the complex point spread function. The proposed imaging system is simple and highly cost-effective. Therefore, it is a promising technique that can be adopted for industrial inspection and security screening.

Ghost translation: an end-to-end ghost imaging approach based on the transformer network

Artificial intelligence has recently been widely used in computational imaging. The deep neural network (DNN) improves the signal-to-noise ratio of the retrieved images, whose quality is otherwise corrupted due to the low sampling ratio or noisy environments. This work proposes a new computational imaging scheme based on the sequence transduction mechanism with the transformer network. The simulation database assists the network in achieving signal translation ability. The experimental single-pixel detector's signal will be 'translated' into a 2D image in an end-to-end manner. High-quality images with no background noise can be retrieved at a sampling ratio as low as 2%. The illumination patterns can be either well-designed speckle patterns for sub-Nyquist imaging or random speckle patterns. Moreover, our method is robust to noise interference. This translation mechanism opens a new direction for DNN-assisted ghost imaging and can be used in various computational imaging scenarios.

Using FADOF to eliminate the background light influence in ghost imaging

The high solar background during the day adversely affects the long distance daytime operations of ghost imaging. It is extremely hard to distinguish the signal light from the background noise light after they are both converted to voltage or current signals by the bucket detector, so spectral filtering before the detector is quite important. In this work, a Faraday anomalous dispersion optical filter (FADOF) is used in eliminating the background light influence in ghost imaging. Results of lab experiment show that the background light noise tolerance of the ghost imaging with FADOF is at least 18 times bigger than that with a 10 nm optical filter. The method has simple structure, great performance and great algorithms compatibility.

Enhanced Sensing Capacity of Terahertz Triple-Band Metamaterials Absorber Based on Pythagorean Fractal Geometry

A new design of a triple band perfect metamaterial absorber based on Pythagorean fractal geometry is proposed and analyzed for terahertz sensing applications. The proposed design showed an enhanced sensing performance and achieved three intensive peaks at 33.93, 36.27, and 38.39 THz, corresponding to the absorptivity of 98.5%, 99.3%, and 99.6%, respectively. Due to the symmetrical nature of the recommended design, the structure exhibited the characteristics of independency on the incident wave angles. Furthermore, a parametric study was performed to show the effects of the change in substrate type, resonator material, and substrate thickness on the absorption spectrum. At a fixed analyte thickness (0.5 μm), the resonance frequency of the design was found to be sensitive to the refractive index of the surrounding medium. The proposed design presented three ultra-sensitive responses of 1730, 1590, and 2050 GHz/RIU with the figure of merit (FoM) of 3.20, 1.54, and 4.28, respectively, when the refractive index was changed from 1.0 to 1.4. Additionally, the metamaterial sensor showed a sensitivity of 1230, 2270, and 1580 GHz/μm at the three resonance frequencies, respectively, when it was utilized for the detection of thickness variation at a fixed analyte refractive index (RI) of 1.4. As long as the RI of the biomedical samples is between 1.3 and 1.4, the proposed sensor can be used for biomedical applications.

Cosinusoidal encoding multiplexed structured illumination multispectral ghost imaging

The information dimension obtained by multispectral ghost imaging is more abundant than in single-band ghost imaging. Existing multispectral ghost imaging systems still meet some shortages, such as complex structure or reconstruction time-consuming. Here, an approach of cosinusoidal encoding multiplexed structured illumination multispectral ghost imaging is proposed. It can capture the multispectral image of the target object within one projection cycle with a single-pixel detector while maintaining high imaging efficiency and low time-consuming. The core of the proposed approach is the employed novel encoding strategy which is apt to decode and reconstruct the multispectral image via the Fourier transform. Specifically, cosinusoidal encoding matrices with specific frequency characteristics are fused with the orthogonal Hadamard basis patterns to form the multiplexed structured illumination patterns. A broadband photomultiplier is employed to collect the backscattered signals of the target object interacted by the corresponding structured illumination. The conventional linear algorithm is applied first to recover the mixed grayscale image of the imaging scene. Given the specific frequency distribution of the constructed cosinusoidal encoding matrices, the mixed grayscale image can be converted to the frequency domain for further decoding processing. Then, the pictures of multiple spectral components can be obtained with some manipulations by applying Fourier transform. A series of numerical simulations and experiments verified our proposed approach. The present cosinusoidal encoding multiplexed structured illumination can also be introduced in many other fields of high-dimensional information acquisition, such as high-resolution imaging and polarization ghost imaging.

Real-time single-pixel imaging using a system on a chip field-programmable gate array

Unlike conventional imaging, the single-pixel imaging technique uses a single-element detector, which enables high sensitivity, broad wavelength, and noise robustness imaging. However, it has several challenges, particularly requiring extensive computations for image reconstruction with high image quality. Therefore, high-performance computers are required for real-time reconstruction with higher image quality. In this study, we developed a compact dedicated computer for single-pixel imaging using a system on a chip field-programmable gate array (FPGA), which enables real-time reconstruction at 40 frames per second with an image size of 128 × 128 pixels. An FPGA circuit was implemented with the proposed reconstruction algorithm to obtain higher image quality by introducing encoding mask pattern optimization. The dedicated computer can accelerate the reconstruction 10 times faster than a recent CPU. Because it is very compact compared with typical computers, it can expand the application of single-pixel imaging to the Internet of Things and outdoor applications.

Deterministic Terahertz Wave Control in Scattering Media

Scattering-assisted synthesis of broadband optical pulses is recognized to have a cross-disciplinary fundamental and application importance. Achieving full-waveform synthesis generally requires means for assessing the instantaneous electric field, i.e., the absolute electromagnetic phase. These are generally not accessible to established methodologies for scattering-assisted pulse envelope and phase shaping. The lack of field sensitivity also results in complex indirect approaches to evaluate the scattering space-time properties. The terahertz frequency domain potentially offers some distinctive new possibilities, thanks to the availability of methods to perform absolute measurements of the scattered electric field, as opposed to optical intensity-based diagnostics. An interesting conceptual question is whether this additional degree of freedom can lead to different types of methodologies toward wave shaping and direct field-waveform control. In this work, we theoretically investigate a deterministic scheme to achieve broadband, spatiotemporal waveform control of terahertz fields mediated by a scattering medium. Direct field access via time-domain spectroscopy enables a process in which the field and scattering matrix of the medium are assessed with minimal experimental efforts. Then, illumination conditions for an arbitrary targeted output field waveform are deterministically retrieved through numerical inversion. In addition, complete field knowledge enables reconstructing field distributions with complex phase profiles, as in the case of phase-only masks and optical vortices, a significantly challenging task for traditional implementations at optical frequencies based on intensity measurements aided with interferometric techniques.

Dual-color terahertz spatial light modulator for single-pixel imaging

Spatial light modulators (SLM), capable of dynamically and spatially manipulating electromagnetic waves, have reshaped modern life in projection display and remote sensing. The progress of SLM will expedite next-generation communication and biomedical imaging in the terahertz (THz) range. However, most current THz SLMs are adapted from optical alternatives that still need improvement in terms of uniformity, speed, and bandwidth. Here, we designed, fabricated, and characterized an 8 × 8 THz SLM based on tunable liquid crystal metamaterial absorbers for THz single-pixel compressive imaging. We demonstrated dual-color compressive sensing (CS) imaging for dispersive objects utilizing the large frequency shift controlled by an external electric field. We developed auto-calibrated compressive sensing (ACS) algorithm to mitigate the impact of the spatially nonuniform THz incident beam and pixel modulation, which significantly improves the fidelity of reconstructed images. Furthermore, the complementary modulation at two absorption frequencies enables Hadamard masks with negative element values to be realized by frequency-switching, thereby halving the imaging time. The demonstrated imaging system paves a new route for THz single-pixel multispectral imaging with high reliability and low cost.

Retina-like Computational Ghost Imaging for an Axially Moving Target

Unlike traditional optical imaging schemes, computational ghost imaging (CGI) provides a way to reconstruct images with the spatial distribution information of illumination patterns and the light intensity collected by a single-pixel detector or bucket detector. Compared with stationary scenes, the relative motion between the target and the imaging system in a dynamic scene causes the degradation of reconstructed images. Therefore, we propose a time-variant retina-like computational ghost imaging method for axially moving targets. The illuminated patterns are specially designed with retina-like structures, and the radii of foveal region can be modified according to the axial movement of target. By using the time-variant retina-like patterns and compressive sensing algorithms, high-quality imaging results are obtained. Experimental verification has shown its effectiveness in improving the reconstruction quality of axially moving targets. The proposed method retains the inherent merits of CGI and provides a useful reference for high-quality GI reconstruction of a moving target.

THz-photonics transceivers by all-dielectric phonon-polariton nonlinear nanoantennas

The THz spectrum (spanning from 0.3 to 30 THz) offers the potential of a plethora of applications, ranging from the imaging through non transparent media to wireless-over-fiber communications and THz-photonics. The latter framework would greatly benefit from the development of optical-to-THz wavelength converters. Exploiting Difference Frequency Generation in a nonlinear all dielectric nanoantenna, we propose a compact solution to this problem. By means of a near-infrared pump beam (at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\omega _1$$\end{document}ω1), the information signal in the optical domain (at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\omega _2$$\end{document}ω2) is converted to the THz band (at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\omega _3=\omega _2-\omega _1$$\end{document}ω3=ω2-ω1). The approach is completely transparent with respect to the modulation format, and can be easily integrated in a metasurface platform for simultaneous frequency and spatial moulding of THz beams.

Spectral imaging of pharmaceutical materials with a compact terahertz difference-frequency generation semiconductor source

Spectral imaging of pharmaceutical material using a compact ultra-broadband (1-4 THz) terahertz semiconductor source was demonstrated. False-color RGB images could be obtained using a simple procedure (calibration free). The ability to distinguish the polymorphism of carbamazepine (CBZ), the hydrate forms of D-(+)-glucose and caffeine, and the crystallinity of nifedipine was demonstrated using the THz DFG source. Crystal forms of pharmaceutical materials can be distinguished using this method.

Multi-Target Tracking Using Windowed Fourier Single-Pixel Imaging

The single-pixel imaging (SPI) technique enables the tracking of moving targets at a high frame rate. However, when extended to the problem of multi-target tracking, there is no effective solution using SPI yet. Thus, a multi-target tracking method using windowed Fourier single-pixel imaging (WFSI) is proposed in this paper. The WFSI technique uses a series of windowed Fourier basis patterns to illuminate the target. This method can estimate the displacements of K independently moving targets by implementing 6K measurements and calculating 2K windowed Fourier coefficients, which is a measurement method with low redundancy. To enhance the capability of the proposed method, we propose a joint estimation approach for multi-target displacement, which solves the problem where different targets in close proximity cannot be distinguished. Using the independent and joint estimation approaches, multi-target tracking can be implemented with WFSI. The accuracy of the proposed multi-target tracking method is verified by numerical simulation to be less than 2 pixels. The tracking effectiveness is analyzed by a video experiment. This method provides, for the first time, an effective idea of multi-target tracking using SPI.

Speckle patterns formed by broadband terahertz radiation and their applications for ghost imaging

Speckle patterns can be very promising for many applications due to their unique properties. This paper presents the possibility of numerically and experimentally formation of speckle patterns using broadband THz radiation. Strong dependence of the statistical parameters of speckles, such as size and sharpness on the parameters of the diffuser are demonstrated: the correlation length and the mean square deviation of the phase surface inhomogeneity. As the surface correlation length is increasing, the speckle size also increases and its sharpness goes down. Alternatively, the magnification of the standard deviation of the surface height leads to the speckle size diminishing and growth of the speckle sharpness. The dimensions of the experimentally formed speckles correspond to the results of numerical simulation. The possibility of utilizing formed speckle patterns for the implementation of the ghost imaging technique has been demonstrated by methods of numerical modeling.

Roadmap of Terahertz Imaging 2021

In this roadmap article, we have focused on the most recent advances in terahertz (THz) imaging with particular attention paid to the optimization and miniaturization of the THz imaging systems. Such systems entail enhanced functionality, reduced power consumption, and increased convenience, thus being geared toward the implementation of THz imaging systems in real operational conditions. The article will touch upon the advanced solid-state-based THz imaging systems, including room temperature THz sensors and arrays, as well as their on-chip integration with diffractive THz optical components. We will cover the current-state of compact room temperature THz emission sources, both optolectronic and electrically driven; particular emphasis is attributed to the beam-forming role in THz imaging, THz holography and spatial filtering, THz nano-imaging, and computational imaging. A number of advanced THz techniques, such as light-field THz imaging, homodyne spectroscopy, and phase sensitive spectrometry, THz modulated continuous wave imaging, room temperature THz frequency combs, and passive THz imaging, as well as the use of artificial intelligence in THz data processing and optics development, will be reviewed. This roadmap presents a structured snapshot of current advances in THz imaging as of 2021 and provides an opinion on contemporary scientific and technological challenges in this field, as well as extrapolations of possible further evolution in THz imaging.

Time division ghost imaging

In the conventional ghost imaging, it requires to illuminate a large number of patterns in order to reconstruct a good quality image under a low signal-to-noise ratio. We propose a new method so called time division ghost imaging to improve the quality of the image in noisy environment. In this procedure, the total number of patterns in the calculation process of the correlation function are divided into the sub-units with fewer illuminated patterns. Then one calculates the correlation for each sub-unit, and synthesizes the intermediate images obtained at each sub-unit. The validation and effectiveness of this method are confirmed by simulation and experiment, showing the robustness to noise.

Improving the performance of ghost imaging via measurement-driven framework

High-quality reconstruction under a low sampling rate is very important for ghost imaging. How to obtain perfect imaging results from the low sampling rate has become a research hotspot in ghost imaging. In this paper, inspired by matrix optimization in compressed sensing, an optimization scheme of speckle patterns via measurement-driven framework is introduced to improve the reconstruction quality of ghost imaging. According to this framework, the sampling matrix and sparse basis are optimized alternately using the sparse coefficient matrix obtained from the low-dimension pseudo-measurement process and the corresponding solution is obtained analytically, respectively. The optimized sampling matrix is then dealt with non-negative constraint and binary quantization. Compared to the developed optimization schemes of speckle patterns, simulation results show that the proposed scheme can achieve better reconstruction quality with the low sampling rate in terms of peak signal-to-noise ratio (PSNR) and mean structural similarity index (MSSIM). In particular, the lowest sampling rate we use to achieve a good performance is about 6.5%. At this sampling rate, the MSSIM and PSNR of the proposed scheme can reach 0.787 and 17.078 dB, respectively.

All-optical spatial terahertz modulator with surface-textured and passivated silicon

For a Si-based all-optical spatial terahertz modulator (STM), an enhanced modulation efficiency under low illumination density would be of great significance to exploit the competence of THz technology in real-world applications. We presented here an implementation of such a device by microtexturing and passivating the Si surface, forming a truncated pyramidal array (TPA). This TPA structure with SiO₂ passivating coatings not only decreases light reflectance and expands the active area for THz modulation but also remarkably increases the photogenerated carrier lifetime. These 3-fold benefits render Si-TPA superior to bare-Si with respect to the achievable modulation efficiency, especially at low irradiation power. Furthermore such a Si-TPA device is also more applicable than its counterpart that is only passivated by SiO₂ nanocoatings, even though the Si-SiO₂ has a slightly increased modulation efficiency. These periodically aligned pyramids resembled as a mesa array significantly suppress the lateral diffusion induced by longer diffusion, resulting in an equivalent resolution of bare-Si. This novel Si-TPA based STM is highly desired for realizing a high-performance THz imager and provides a feasible approach to breaking the trade-off between resolution and modulation efficiency.

All-Optical Two-Color Terahertz Emission from Quasi-2D Nonlinear Surfaces

Two-color terahertz (THz) generation is a field-matter process combining an optical pulse and its second harmonic. Its application in condensed matter is challenged by the lack of phase matching among multiple interacting fields. Here, we demonstrate phase-matching-free two-color THz conversion in condensed matter by introducing a highly resonant absorptive system. The generation is driven by a third-order nonlinear interaction localized at the surface of a narrow-band-gap semiconductor, and depends directly on the relative phase between the two colors. We show how to isolate the third-order effect among other competitive THz-emitting surface mechanisms, exposing the general features of the two-color process.

Denoising ghost imaging under a small sampling rate via deep learning for tracking and imaging moving objects

Ghost imaging (GI) usually requires a large number of samplings, which limit the performance especially when dealing with moving objects. We investigated a deep learning method for GI, and the results show that it can enhance the quality of images with the sampling rate even down to 3.7%. With a convolutional denoising auto-encoder network trained with numerical data, blurry images from few samplings can be denoised. Then those outputs are used to reconstruct both the trajectory and clear image of the moving object via cross-correlation based GI, with the number of required samplings reduced by two-thirds.

Single-pixel imaging using a recurrent neural network combined with convolutional layers

Single-pixel imaging allows for high-speed imaging, miniaturization of optical systems, and imaging over a broad wavelength range, which is difficult by conventional imaging sensors, such as pixel arrays. However, a challenge in single-pixel imaging is low image quality in the presence of undersampling. Deep learning is an effective method for solving this challenge; however, a large amount of memory is required for the internal parameters. In this study, we propose single-pixel imaging based on a recurrent neural network. The proposed approach succeeds in reducing the internal parameters, reconstructing images with higher quality, and showing robustness to noise.

Computational ghost imaging with spatiotemporal encoding pseudo-random binary patterns

Computational ghost imaging (CGI) can reconstruct the pixelated image of a target without lenses and image sensors. In almost all spatial CGI systems using various patterns reported in the past, people often only focus on the distribution of patterns in the spatial dimension but ignore the possibility of encoding in the time dimension or even the space-time dimension. Although the random illumination pattern in CGI always brings some inevitable background noise to the recovered image, it has considerable advantages in optical encryption, authentication, and watermarking technologies. In this paper, we focus on stimulating the potential of random lighting patterns in the space-time dimension for embedding large amounts of information. Inspired by binary CGI and second-order correlation operations, we design two novel generation schemes of pseudo-random patterns for information embedding that are suitable for different scenarios. Specifically, we embed a total of 10,000 ghost images (64 × 64 pixels) of the designed Hadamard-matrix-based data container patterns in the framework of CGI, and these ghost images can be quantitatively decoded to two 8-bit standard grayscale images, with a total data volume of 1, 280, 000 bits. Our scheme has good noise resistance and a low symbol error rate. One can design the number of lighting patterns and the information capacity of the design patterns according to the trade-off between accuracy and efficiency. Our scheme, therefore, paves the way for CGI using random lighting patterns to embed large amounts of information and provides new insights into CGI-based encryption, authentication, and watermarking technologies.

Image recovery of ghost imaging with sparse spatial frequencies

When the spatial frequencies of the object are insufficiently sampled, the reconstruction of ghost imaging will suffer from repetitive visual artifacts, which cannot be effectively tackled by existing ghost imaging reconstruction techniques. In this Letter, extensions of the CLEAN algorithm applied in ghost imaging are explored to eliminate those artifacts. Combined with the point spread function estimation using the second-order coherence measurement in ghost imaging, our modified CLEAN algorithm is demonstrated to have a fast and noteworthy improvement against the spatial-frequency insufficiency, even for the extreme sparse sampling cases. A brief explanation of the algorithm and performance analysis are given.

Broadband high-resolution terahertz single-pixel imaging

We report a simple single-pixel imaging system with a low mean squared error in the entire terahertz frequency region (3-13 THz) that employs a thin metallic ring with a series of directly perforated random masks and a subpixel mask digitization technique. This imaging system produces high pixel resolution reconstructed images, up to 1200 × 1200 pixels, and imaging area of 32 × 32 mm². It can be extended to develop advanced imaging systems in the near-ultraviolet to terahertz region.

Ghost polarimetry with unpolarized pseudo-thermal light

We present an experimental implementation of the ghost polarimetry concept in unpolarized light, which allows obtaining complete information on the spatial distribution of polarization properties of objects with linear dichroism. It is theoretically shown that it is possible to restore the spatial distribution of the azimuth and a value of anisotropy of such objects. The developed technique allows us to free up the object arm from all additional optical elements, including polarizers. The experimental results of measuring the dichroism parameters of a test four-sectional sample are presented, which demonstrate the efficiency of the method and confirm the correctness of the developed theoretical model.