To address the issue of iterative stagnation caused by twin-image artifacts in conventional phase retrieval methods based on global support constraints, an adaptive partitioned support constraint (APS) of the phase retrieval algorithm utilizing a discrimination of correlation coefficient threshold is proposed in this paper. The algorithm calculates the correlation coefficient δ between the Fourier amplitude obtained through iterative updates and the target Fourier amplitude obtained from the speckle autocorrelation. When the correlation coefficient δ is below the presupposed threshold, the processing mechanism of support constraint divided into two pieces is automatically triggered. Numerical simulations demonstrate that, compared to conventional methods, the phase retrieval algorithm of adaptive partitioned support constraint dramatically mitigates the impact of twin images, showing advantages by both the sum-squared error (SSE) and the image universal quality index (UQI) evaluation metrics. The phase retrieval algorithm of adaptive partitioned support constraint mitigates the interference of twin images during the iteration process, providing new, to our knowledge, insights into the speckle autocorrelation imaging and promoting practical applications of scattering imaging technology in fields such as biomedicine and deep-sea exploration. Furthermore, this algorithm can be combined with X-ray imaging and Fourier ptychographic microscopy (FPM) technology to reduce imaging time and improve imaging quality.

Objective To meet the demands of exploration and scientific research on extraterrestrial bodies, acquiring crater detection information and cataloged data is essential as the foundation for measurement and analysis. Although remote sensing payloads capture data containing craters, these images feature a large number of targets, high target density, and diverse morphologies. Detecting craters in collected image data is essential for further scientific research, measurement, and localization on celestial surfaces. The discovery of new craters and the establishment of a complete crater cataloging database are prerequisites for the study of craters and related downstream tasks. Existing detected and cataloged celestial crater datasets cover only parts of the Moon and Martian surfaces. With the advancement of future deep-space exploration, artificial intelligence methodologies are anticipated to supersede conventional manual identification approaches and emerge as the predominant method for celestial crater detection. However, AI-driven approaches depend on well-annotated training datasets, and manual annotation poses inherent challenges such as high labor intensity and technical complexity. This imperative has propelled simulation-based methodologies for generating synthetic remote sensing imagery of craters into a critical research focus for constructing training datasets. Methods In this paper, we propose an efficient crater image simulation method based on neural radiance field (NeRF). The method integrates image simulation technology, embedding the target 3D model into the process to simulate and render crater images with varied morphologies and illumination conditions. It implicitly captures the geometric structure of the target in the model and generates images that approximate the physical imaging process by incorporating parameters such as imaging distance, angle, and light angle. The method consists of two parts: NeRF-based image generation and image-harmonization-based target fusion. First, the 3D model is used to construct the crater data and train the NeRF model for crater image generation. Then, the imaging parameters of the remote sensing image are fed into the NeRF network to generate crater images. Next, the feature domain difference between the crater target and the background is adjusted by constructing and training an image harmonization network in combination with the NeRF algorithm. This approach compensates for fringe distortions between the target and background while simultaneously producing positional labels for the crater. The proposed method fulfills simulation requirements according to specific crater morphological types and actual imaging conditions, ensuring both scientific accuracy in crater representation and visual consistency with realistic celestial surface environments. Results and Discussions The simulated crater largely matches the geometric structure of actual craters, though some differences remain compared with real images (Fig. 6). The harmonized crater aligns more closely with the background's feature domain while preserving its geometric structure. In this paper, the effectiveness of the proposed simulation method is verified by the target detection algorithm. Using our annotated lunar crater dataset, experiments show that the proposed method, leveraging small-scale real image datasets, produces effective, controllable training data for multiple crater detection approaches. As shown in Tables 1-3, adding crater images generated by our method improves detection metrics in every group, with a maximum gain of 27.2 degrees o and an average F1-score improvement of 11.3 degrees o. It is demonstrated that the crater simulation method can provide augmented data for many types of target detection algorithms, improve target detection accuracy, and remain applicable to different datasets. Moreover, our simulated images outperform those generated by three other mainstream image simulation algorithms. The results substantiate that our approach effectively supports the training of deep learning-based crater detection methods. Conclusions To address the bottleneck of insufficient training data for deep-learning-based crater detection algorithms, we propose a remote sensing image simulation method of craters with neural radiance fields. The method integrates image fusion and harmonization techniques, coupling a crater 3D model with the image simulation algorithm, allowing control over target illumination and imaging conditions, and generating crater simulation data under complex and varied scenarios. Experimental results show that the simulated images expand training datasets, improve detection accuracy, and support high-precision mapping of celestial topography. This simulation method can also be used for simulated image generation in remote sensing detection-related fields, providing diverse training data and annotations for multiple target types, meeting data volume requirements for high-precision detection, recognition, semantic segmentation, and related tasks.

Optical neural networks present distinct advantages over traditional electrical counterparts, such as accelerated data processing and reduced energy consumption. While coherent light is conventionally used in optical neural networks, our study proposed harnessing spatially incoherent light in all-optical Fourier neural networks. Contrary to natural predictions of declining target recognition accuracy with increased incoherence, our experimental results demonstrated a surprising outcome: improved accuracy with incoherent light. We attribute this enhancement to spatially incoherent light's ability to alleviate experimental errors like diffraction rings and laser speckle. Our experiments introduced controllable spatial incoherence by passing monochromatic light through a spatial light modulator featuring a dynamically changing random phase array. These findings underscore partially coherent light's potential to optimize optical neural networks, delivering dependable and efficient solutions for applications demanding consistent accuracy and robustness across diverse conditions, including on-chip optical computing, photonic interconnects, and reconfigurable optical processors.

Video snapshot compressive imaging(Video SCI) modulates scenes using various encoding masks and captures compressed measurements with a low-speed camera during a single exposure. Subsequently, reconstruction algorithms restore image sequences of dynamic scenes, offering advantages such as reduced bandwidth and storage space requirements. The temporal correlation in video data is crucial for Video SCI, as it leverages the temporal relationships among frames to enhance the efficiency and quality of reconstruction algorithms, particularly for fast-moving objects.This paper discretizes video frames to create image datasets with the same data volume but differing temporal correlations. We utilized the state-of-the-art(SOTA) reconstruction framework, EfficientSCI++, to train various compressed reconstruction models with these differing temporal correlations. Evaluating the reconstruction results from these models, our simulation experiments confirm that a reduction in temporal correlation leads to decreased reconstruction accuracy. Additionally, we simulated the reconstruction outcomes of datasets devoid of temporal correlation, illustrating that models trained on non-temporal data affect the temporal feature extraction capabilities of transformers, resulting in negligible impacts on the evaluation of reconstruction results for non-temporal correlation test datasets.

Objective Femtosecond laser pulse underwater filamentation transmission and acoustic wave excitation research represents a significant direction in the field of underwater detection. This study systematically explores the transmission characteristics of femtosecond lasers in aqueous media and the mechanisms of acoustic wave excitation by establishing a comprehensive theoretical model system. The research employs the nonlinear Schrodinger equation to describe the evolution of the laser electric field envelope while constructing an electron density equation that incorporates multiphoton ionization and avalanche ionization mechanisms, providing a theoretical foundation for understanding underwater filament formation. Methods In terms of research methodology, this study innovatively combines theoretical modeling with numerical simulation. Through the standard Fourier. Crank. Nicholson numerical solution method, high-precision simulation of the femtosecond laser underwater transmission process was achieved. Regarding simulation parameter settings, the transverse spatial resolution was controlled at 4.39 mu m (less than 10 mu m), and the temporal resolution reached 0.53 fs (less than 3 fs), ensuring the reliability of the computational results. The study focused on examining the effects of group velocity dispersion and seawater attenuation on the transmission process, providing important references for practical applications. Results and Discussions The research results demonstrate that laser pulse parameters have a decisive influence on underwater transmission performance. Using long-pulse-width laser pulses can effectively suppress optical field diffusion and significantly improve transmission stability. When the focal length increases from 5 m to 10 m, the peak plasma density can be increased to the order of 1016 cm(-3), with the light intensity exceeding the clamping intensity in water ( approximately 2.18x10(11) W/cm(2)). This discovery provides clear guidance for optimizing laser system parameters. In terms of wavelength selection, 532 nm lasers demonstrate significant advantages. Compared to the 800 nm lasers, 532 nm lasers exhibit higher multiphoton ionization rate ( K=3) and lower attenuation coefficient. Under the same conditions, the peak plasma density generated by 532 nm lasers can reach 3.2x1018 cm(-3), with an acoustic pressure amplitude of 1.2x10(6) Pa, approximately 2.3 and 2.4 times higher than those of 800 nm lasers, respectively. These results provide important evidence for laser wavelength selection. To address the attenuation problem in actual marine environments, the study proposes effective solutions. Under Jerlov Type 3 turbid water conditions (attenuation coefficient C= 0.595 m(-1)), reducing the focal distance to 15. 25 cm can achieve a deposited energy density of 5.75 mJ/ mm3 and excite an acoustic pressure of 1.9x10(6) Pa. The optimized parameter combination can maintain sufficient energy deposition in high-attenuation environments, ensuring effective acoustic wave excitation. Regarding the photoacoustic conversion mechanism, a complete theoretical model was established. By calculating the average energy deposition in the focal region, the expression for energy deposition per unit volume was derived. Based on thermodynamic principles, a mathematical model for acoustic pressure generation in laser-heated regions was developed. The research shows that at a focal length of 25 cm, the maximum temperature difference in the medium surrounding the filament is 2.7 K, corresponding to an acoustic pressure of 1.4x10(5) Pa. When the focal length is reduced to 15 cm, the temperature difference increases to 3.8 K, and the acoustic pressure rises to 1.9x10(6) Pa, representing an increase of 13 times. These results provide quantitative evidence for understanding the photoacoustic conversion mechanism. Conclusions This study not only deepens the understanding of the physical processes involved in femtosecond laser underwater transmission but also provides important references for developing new underwater detection technologies. The research findings have broad application prospects in fields such as marine resource exploration, underwater target identification, and environmental monitoring. Future research could further explore transmission laws in dynamic marine environments, optimize system parameters, and promote the practical application of this technology.

Physics-informed neural networks (PINN) have achieved notable success in solving partial differential equations (PDE), yet solving the Navier-Stokes equations (NSE) with complex boundary conditions remains a challenging task. In this paper, we introduce a novel Hybrid Boundary PINN (HB-PINN) method that combines a pretrained network for efficient initialization with a boundary-constrained mechanism. The HB-PINN method features a primary network focused on inner domain points and a distance metric network that enhances predictions at the boundaries, ensuring accurate solutions for both boundary and interior regions. Comprehensive experiments have been conducted on the NSE under complex boundary conditions, including the 2D cylinder wake flow and the 2D blocked cavity flow with a segmented inlet. The proposed method achieves state-of-the-art (SOTA) performance on these benchmark scenarios, demonstrating significantly improved accuracy over existing PINN-based approaches.

With the exponential advancement of high-resolution spaceborne optical remote sensing technologies, satellite systems have achieved a remarkable transition from meter-level to sub-meter spatial resolutions, resulting in daily data generation volumes exceeding terabytes. However, wide Field-of-View (FoV) optical architectures-such as off-axis four-mirror configurations with FoV angles exceeding 30 degrees-inherently suffer from edge defocus blur due to spherical aberrations and the mismatch between curved focal surfaces and planar imaging sensors. This spatially non-uniform optical degradation critically impairs the performance of Compressed Sensing (CS)-based Snapshot Compressive Imaging (SCI) systems, which rely on the accurate reconstruction of sparse signals from under-sampled measurements. Although Optical Neural Networks (ONNs) have demonstrated exceptional dimensionality reduction capabilities (up to 90%) through in-sensor computation, they often fail to preserve raw light-field information necessary for quantitative inversion and downstream multi-task analysis. In response to these limitations, this study systematically investigates the nonlinear interactions between edge defocus and Block-based Compressed Sensing (BCS) reconstruction performance. A novel spatially adaptive modeling and processing framework is proposed to guide imaging system design and CS parameter optimization under wide-FoV constraints. In wide-angle optical systems, defocus blur exhibits inherent rotational symmetry, reflecting the geometric characteristics of spherical optical surfaces. To model this Spatially Varying Point Spread Function (SVPSF), we introduce a Gaussian Kernel Mixture (GKM) approach, which partitions the image plane into concentric annular regions. Each region is assigned an independent Gaussian kernel characterized by location-dependent standard deviation (sigma) and kernel size K (ksize). The PSF at any spatial coordinate (m,n) is expressed as b(( m,n )) = & sum;(K)(k = 1) beta(k)& sdot;G(x,y|sigma(k)), where beta(k )are learnable mixture weights optimized through iterative deblurring strategies inspired by fixed-point theory. Optical simulations conducted in ZEMAX confirmed the rotational symmetry of defocused PSFs, while BRISQUE-based no-reference quality assessments further validated the spatially non-uniform degradation characteristics, with central regions exhibiting scores of 15.65 compared to 52.46 in peripheral areas. To simulate realistic spaceborne imagery, we applied the proposed SVPSF modeling technique to high-resolution remote sensing images from the DOTA dataset. Each image was convolved with a region-specific GKM kernel, introducing radially increasing blur to replicate defocus effects. A modular image processing pipeline was constructed, comprising three key stages: 1) optical degradation modeling via SVPSF convolution, 2) block-wise CS encoding using binary random masks, and 3) image reconstruction using a Deep Unfolding Network (DUN). The DUN employed gated 3D convolution layers alongside a Two-Way Cross-Attention (TWCA) mechanism to effectively mitigate block artifacts and enhance spatial coherence in reconstructed images. Experimental evaluation revealed a nonlinear relationship between blur parameters (sigma and ksize) and image reconstruction performance. When using a 3x3 kernel, increasing sigma resulted in a 5.31 dB decline in PSNR, from 33.99 dB to 28.68 dB, and a modest SSIM reduction of 0.077 4 (from 0.871 to 0.793 6). Enlarging the kernel to sizes such as 13x13 led to only marginal PSNR degradation (<0.18 dB) but caused accelerated structural degradation, with SSIM declining from 0.83 to 0.797. Interestingly, an intermediate blur level (sigma =1.5) yielded the most favorable balance between detail preservation and noise suppression, suggesting an optimal operating condition for BCS systems under moderate defocus. To quantify the constraints of system design, we constructed a triadic coupling model describing the interaction between FoV-induced blur, CS Compression Ratio (CR), and reconstruction fidelity. The results showed that to maintain an acceptable PSNR threshold of >= 30 dB, the compression ratio must not exceed 4 when the blur parameter sigma remains <= 2.5. At higher CR values (e.g., CR=16), reconstruction quality deteriorated rapidly, with PSNR falling from 28.33 dB to 23.12 dB as defocus increased. Notably, peripheral image regions suffered from a 62.4% loss in texture signal-to-noise ratio compared to the image center, emphasizing the necessity for spatially adaptive CS strategies to account for heterogeneous image degradations. This work offers several key technical contributions. First, we introduce a GKM-based SVPSF modeling framework that accurately simulates edge defocus characteristics with 92% fidelity, as validated through optical simulations and perceptual quality metrics. Second, we propose an enhanced DUN reconstruction architecture incorporating a TWCA attention mechanism, which reduces block artifacts by 34% relative to traditional 3D U-Net architectures. Third, we establish quantitative design guidelines for spaceborne imaging systems by mapping the relationship between FoV, compression ratio, and reconstruction accuracy. These insights facilitate informed decisions in payload design, especially for systems constrained by bandwidth, resolution, and energy budgets. The proposed framework has wide-reaching implications for both theoretical research and practical deployment. From a hardware design perspective, our findings inform the optimization of optical payloads, particularly the trade-off between wide FoV and PSF uniformity, which can be mitigated through aspheric element design. In terms of onboard processing, the modular reconstruction pipeline enables real-time reconstruction under strict power constraints, which is critical for micro-and nano-satellite missions.

Video Snapshot Compressive Imaging (SCI) aims to capture high-speed scenes with low-speed cameras in a lowcost and low-bandwidth manner. Specifically, a high-speed scene is encoded by different modulation masks and then summed up to generate a snapshot compressed measurement which is finally captured by a traditional lowspeed camera. Following this, reconstruction algorithms are correspondingly designed to retrieve the compressed dynamic scene. Existing video SCI reconstruction algorithms have achieved superior performance in the simulated testing data and real testing data with less noise. However, in the applications of real imaging systems, the existence of the intrinsic noise within detector results in the mismatch between the simulated and real imaging systems. Therefore, intrinsic noise and light intensity become the major challenges for video SCI reconstruction in the real cases. Bearing the above in mind, in this paper, we propose to integrate the intrinsic noise of the real imaging system into the whole reconstruction pipeline. More importantly, based on the noise-integrated framework, we evaluate the reconstruction performance under different light conditions and compression ratios. Experimental results show that with different signal-to-noise ratios, there exists an extreme performance bound that is lower than that of the noise-free condition. To verify the effectiveness of our proposed method, we build a real video SCI system and carefully calibrate its intrinsic noise. Following this, existing state-of-the-art reconstruction method EfficientSCI is used to present the reconstruction results. Introducing calibrated intrinsic noise significantly improves reconstruction quality under noisy and insufficient-light conditions, bringing performance close to that of a noise-free scenario. The proposed method has further been verified by the experimental results.

All-optical neural networks (AONNs) have emerged as a promising paradigm for ultrafast and energy-efficient computation. These networks typically consist of multiple serially connected layers between input and output layers–a configuration we term spatially series AONNs, with deep neural networks (DNNs) being the most prominent examples. However, such series architectures suffer from progressive signal degradation during information propagation and critically require additional nonlinearity designs to model complex relationships effectively. Here we propose a spatially parallel architecture for all-optical neural networks (SP-AONNs). Unlike series architecture that sequentially processes information through consecutively connected optical layers, SP-AONNs divide the input signal into identical copies fed simultaneously into separate optical layers. Through coherent interference between these parallel linear sub-networks, SP-AONNs inherently enable nonlinear computation without relying on active nonlinear components or iterative updates. We implemented a modular 4F optical system for SP-AONNs and evaluated its performance across multiple image classification benchmarks. Experimental results demonstrate that increasing the number of parallel sub-networks consistently enhances accuracy, improves noise robustness, and expands model expressivity. Our findings highlight spatial parallelism as a practical and scalable strategy for advancing the capabilities of optical neural computing.

The plaque biofilm is a structurally diverse and compositionally complex aggregate, serving as the initiating factor of peri-implantitis and is closely related to the onset and progression of this disease (BMC Oral Health 24:105, 2024), (NPJ Biofilms Microbiomes 10:12, 2024). Previous studies have shown that compared to the microbial communities in healthy peri-implant tissues, the microbial communities beneath the mucosa in peri-implantitis are more complex and diverse, with a significant increase in periodontal pathogens. Due to the lack of cementum and periodontal ligament as a protective system, dental implants are more susceptible to bacterial infections than natural teeth. Therefore, the elimination of plaque biofilm is crucial in the treatment of peri-implantitis; the biofilm formed after implant debridement may affect the treatment outcome and the long-term stability of the implant. To explore these two issues, we conducted the following experiment. This experiment comparatively observed the microstructural changes on the surfaces of commonly used implant materials after traditional and laser treatments, evaluated their decontamination capabilities and antibacterial effects, thereby providing a theoretical basis for the use of lasers in the prevention and treatment of peri-implantitis. The experiment is divided into 5 groups: normal group (A), carbon fiber scraping group (B), sandblasting group (C), Er: YAG laser group (E), and Nd: YAG laser group (N).Use a K-type thermocouple temperature measuring instrument to monitor the temperature changes on the material surface during processing; use a roughness measuring instrument to measure the surface roughness (Ra) of the material after processing. Plaque biofilm is formed on the surface of the material. After being treated in different ways, the decontamination ability is tested by scanning electron microscopy observation, live/dead bacterial staining combined with laser confocal imaging system observation and plate colony counting method. After using the above 5 methods to treat the specimen for 60 s, wear it in the mouth of the subject for 24 h, and observe the adhesion of live/dead bacteria with a laser confocal imaging system using live/dead bacterial staining. SPSS 27.0 software was used to perform statistical analysis on each group of data. 1. The heat generated by the four methods during the 60-second treatment process is all within the safe range ( P < 0.05). Carbon fiber scraping treatment has statistically significant changes in the surface roughness of the two materials ( P < 0.05). 2. After 48 hours of SEM observation, the surfaces of the two materials were almost completely covered by plaque biofilm, and the thickness of the plaques varied; the plaques in the three groups C, E, and N were scattered, and the bacterial cell membranes were incomplete, the bacterial cells were broken, and the bacteria were lost. Normal form. The plate colony count showed that the remaining bacterial content on the surface of the two materials in groups B, C, E, and N was much smaller than that in group A (P<0.05). Images obtained by laser confocal microscopy showed that the green fluorescence and red fluorescence of each group B, C, E, and N were significantly reduced, and the average fluorescence intensity was also reduced (P<0.05). 3. Laser scanning confocal microscope images show that each group has a large amount of green fluorescence covering most areas of the specimen surface uniformly or unevenly; the three treatments of C, E, and N can reduce the total average fluorescence intensity of groups Z and T, and The adhesion of dead bacteria on the surface increased (P<0.05). Er:YAG laser and water mist Nd:YAG laser will not have any adverse effects on the surface morphology, roughness and temperature changes of pure titanium and zirconia sheets. The laser is safer to use and can efficiently remove the material surface. Er:YAG laser and water mist Nd:YAG laser irradiation can increase the proportion of dead bacteria in reattached plaque.

Optical Kerr effect, in which input light intensity linearly alters the refractive index, has enabled the generation of optical solitons, supercontinuum spectra, and frequency combs, playing vital roles in the on-chip devices, fiber communications, and quantum manipulations. Especially, terahertz Kerr effect, featuring fascinating prospects in future high-rate computing, artificial intelligence, and cloud-based technologies, encounters a great challenge due to the rather low power density and feeble Kerr response. Here, we demonstrate a giant terahertz frequency Kerr nonlinearity mediated by stimulated phonon polaritons. Under the influences of the giant Kerr nonlinearity, the power-dependent refractive index change would result in a frequency shift in the microcavity, which was experimentally demonstrated via the measurement of the resonant mode of a chip-scale lithium niobate Fabry-Pérot microcavity. Attributed to the existence of stimulated phonon polaritons, the nonlinear coefficient extracted from the frequency shifts is orders of magnitude larger than that of visible and infrared light, which is also theoretically demonstrated by nonlinear Huang equations. This work opens an avenue for many rich and fruitful terahertz Kerr effect based physical, chemical, and biological systems that have terahertz fingerprints.

Soft- and hard-constrained Physics Informed Neural Networks (PINNs) have achieved great success in solving partial differential equations (PDEs). However, these methods still face great challenges when solving the Navier-Stokes equations (NSEs) with complex boundary conditions. To address these challenges, this paper introduces a novel complementary scheme combining soft and hard constraint PINN methods. The soft-constrained part is thus formulated to obtain the preliminary results with a lighter training burden, upon which refined results are then achieved using a more sophisticated hard-constrained mechanism with a primary network and a distance metric network. Specifically, the soft-constrained part focuses on boundary points, while the primary network emphasizes inner domain points, primarily through PDE loss. Additionally, the novel distance metric network is proposed to predict the power function of the distance from a point to the boundaries, which serves as the weighting factor for the first two components. This approach ensures accurate predictions for both boundary and inner domain areas. The effectiveness of the proposed method on the NSEs problem with complex boundary conditions is demonstrated by solving a 2D cylinder wake problem and a 2D blocked cavity flow with a segmented inlet problem, achieving significantly higher accuracy compared to traditional soft- and hard-constrained PINN approaches. Given PINN's inherent advantages in solving the inverse and the large-scale problems, which are challenging for traditional computational fluid dynamics (CFD) methods, this approach holds promise for the inverse design of required flow fields by specifically-designed boundary conditions and the reconstruction of large-scale flow fields by adding a limited number of training input points. The code for our approach will be made publicly available.

Compared to traditional neural networks, optical neural networks demonstrate significant advantages in terms of information processing speed, energy efficiency, anti -interference capability, and scalability. Despite the rapid development of optical neural networks in recent years, most existing systems still face challenges such as complex structures, time-consuming training, and insufficient accuracy. This study fully leverages the coherence of optical systems and introduces an optical Fourier convolutional neural network based on the diffraction of complex image light fields. This new network is not only structurally simple and fast in computation but also excels in image classification accuracy. Our research opens new perspectives for the development of optical neural networks, and also offers insights for future applications in high -efficiency, low -energy -consumption computing domains.

With the advantages of low latency, low power consumption, and high parallelism, optical neural networks (ONN) offer a promising solution for time-sensitive and resource-limited artificial intelligence applications. However, the performance of the ONN model is often diminished by the gap between the ideal simulated system and the actual physical system. To bridge the gap, this work conducts extensive experiments to investigate systematic errors in the optical physical system within the context of image classification tasks. Through our investigation, two quantifiable errors—light source instability and exposure time mismatches—significantly impact the prediction performance of ONN. To address these systematic errors, a physics-constrained ONN learning framework is constructed, including a well designed loss function to mitigate the effect of light fluctuations, a CCD adjustment strategy to alleviate the effects of exposure time mismatches and a ’physics-prior based’ error compensation network to manage other systematic errors, ensuring consistent light intensity across experimental results and simulations. In our experiments, the proposed method achieved a test classification accuracy of 96.5% on the MNIST dataset, a substantial improvement over the 61.6% achieved with the original ONN. For the more challenging QuickDraw16 and Fashion MNIST datasets, experimental accuracy improved from 63.0% to 85.7% and from 56.2% to 77.5%, respectively. Moreover, the comparison results further demonstrate the effectiveness of the proposed physics-constrained ONN learning framework over state-of-the-art ONN approaches. This lays the groundwork for more robust and precise optical computing applications.

Optical remote sensing payloads of high resolution and large field of view are facing the problem of focal plane splicing, big data acquisition and transmission. This paper proposes a novel regime of spatial compressive remote sensing (SCRS) based on snapshot compressive imaging (SCI) technology, which reduces the amount of acquired data via an encoded compressive imaging system while maintains a satisfactory image quality by applying deep learning based reconstruction algorithms. The proposed spatial compressive imaging scheme encodes the images of different fields of view, focuses them on the same detector, and reconstructs the images with an iterative algorithm of a denoising network and projection constraints. The influence of the variations in the compression ratio on the reconstructed image quality is studied by simulation. In addition to conventional image reconstruction evaluation metrics like peak signal-to-noise ratio (PSNR) and structural similarity (SSIM), the target detection accuracy is proposed as an important index for SCRS system performance evaluation. By the Plug-and-Play FFDNet reconstruction network, the compression ratio of SCRS can reach up to 25, with a high average PSNR up to 32 dB and a high average SSIM up to 0.82. The limited decrease in target detection performance shows that SCRS has hopeful prospects in future remote sensing application scenarios.

Low-light image enhancement (LLIE) investigates how to improve the brightness of an image captured in illumination-insufficient environments. The majority of existing methods enhance low-light images in a global and uniform manner, without taking into account the semantic information of different regions. Consequently, a network may easily deviate from the original color of local regions. To address this issue, we propose a semantic-aware knowledge-guided framework (SKF) that can assist a low-light enhancement model in learning rich and diverse priors encapsulated in a semantic segmentation model. We concentrate on incorporating semantic knowledge from three key aspects: a semantic-aware embedding module that adaptively integrates semantic priors in feature representation space, a semantic-guided color histogram loss that preserves color consistency of various instances, and a semantic-guided adversarial loss that produces more natural textures by semantic priors. Our SKF is appealing in acting as a general framework in the LLIE task. We further present a refined framework SKF++ with two new techniques: (a) Extra convolutional branch for intra-class illumination and color recovery through extracting local information and (b) Equalization-based histogram transformation for contrast enhancement and high dynamic range adjustment. Extensive experiments on various benchmarks of LLIE task and other image processing tasks show that models equipped with the SKF/SKF++ significantly outperform the baselines and our SKF/SKF++ generalizes to different models and scenes well. Besides, the potential benefits of our method in face detection and semantic segmentation in low-light conditions are discussed.

- We theoretically propose an efficient method to generate near-circularly polarized isolated attosecond (as) pulses (NCP-IAPs) from a current-carrying state of Ar atom driven by two-color cross-linearly polarized laser fields. We find that the ellipticity of high harmonics can be controlled by adjusting the crossing angle of two linearly polarized lasers and the near-circularly polarized supercontinuum harmonics are obtained when the crossing angle is around 140 degrees. Furthermore, we can produce the NCP-IAPs with the ellipticity up to 0.94 and the shortest one achieves 196 as. This work demonstrates the possibility for generating the NCP-IAPs using a current-carrying state of atoms driven by two-color cross-linearly polarized laser fields.

Holography is an essential technique of generating three-dimensional images. Recently, quantum holography with undetected photons (QHUP) has emerged as a groundbreaking method capable of capturing complex amplitude images. Despite its potential, the practical application of QHUP has been limited by susceptibility to phase disturbances, low interference visibility, and limited spatial resolution. Deep learning, recognized for its ability in processing complex data, holds significant promise in addressing these challenges. In this report, we present an ample advancement in QHUP achieved by harnessing the power of deep learning to extract images from single-shot holograms, resulting in vastly reduced noise and distortion, alongside a notable enhancement in spatial resolution. The proposed and demonstrated deep learning QHUP (DL-QHUP) methodology offers a transformative solution by delivering high-speed imaging, improved spatial resolution, and superior noise resilience, making it suitable for diverse applications across an array of research fields stretching from biomedical imaging to remote sensing. DL-QHUP signifies a crucial leap forward in the realm of holography, demonstrating its immense potential to revolutionize imaging capabilities and pave the way for advancements in various scientific disciplines. The integration of DL-QHUP promises to unlock new possibilities in imaging applications, transcending existing limitations and offering unparalleled performance in challenging environments.

Objective With the development of ultrashort and ultra-intense laser, it has been revealed that when a femtosecond laser pulse propagates in air, filaments, referred to as "filament laser", would occur owing to nonlinear effects. Traditional spaceborne air-pollution monitoring devices rely on spectral imaging and LiDAR technology, which cannot realize real-time monitoring of atmospheric multi-component pollutants, identify unknown pollutants, and detect the chemical composition of various pollutants. The filament laser system in orbit emits a femtosecond laser pulse into the atmosphere, and the intensity of the femtosecond laser pulse is sufficient to ionize molecules in the atmospheric environment. Ionization excites the fluorescence spectrum carrying the information of the material composition, which can determine the various material components, species, and content in the area of the filament laser. To aid the research on space-based filament lidar technology, this study explored the optical system design of a space-based filament lidar spectrometer for remote sensing applications and realized the optical system configuration design. The spectral range and resolution of the spectrometer are 320-950 nm and 2 nm, respectively, and it has applicability in high resolution spectral detection of atmospheric pollutant composition. Methods First, the application requirements of space-based filament lidar spectrometer were analyzed. On the basis of the characteristics of pollutants and the corresponding spectra of the substance elements, the working spectrum of the spectrometer and the spectral resolution were designed to be 320-950 nm and 2 nm, respectively. Using filament laser propagation simulation software, the filament laser diameter was found to be approximately 6 mm after 400 km orbital propagation. The filament laser diameter can be constrained to a small spatial scale after ultralong-distance propagation. To conform to the spectral range and resolution requirements, the size of CCD detector is 1024x1024, with 13 mu mx13 mu m pixel size. The minimum spectral sampling interval was designed to be 0.67 nm/pixel. Considering the signal-to-noise ratio requirements of the spectrometer, the relative aperture of the optical system was determined as D/f ' = 1/3.5, and the aperture of the spectrometer system was set as 0.5 m. Then, considering the requirements of engineering and the space environment, the optical design and optical-mechanical design of the spectrometer were performed so as to provide an effective load scheme for space-based filament laser atmospheric detection. Results and Discussions The optical system of the spectrometer mainly comprises a telescopic system, slit, collimating system, plane grating, and imaging system; the collimating system, dispersion element, and imaging system constitute the spectrometer. The front telescopic system adopts a total reflection Cassegrain structure without chromatic aberration correction, and the root mean square (RMS) value of the diffuse spot radius of its imaging point is within 14 mu m. The spectrometer uses a reflective plane grating; the number of plane grating lines was determined to be approximately 263 lp/mm, and the grating aperture was 22 mm. A spectral resolution of 2 nm was achieved using first-order diffraction light. The maximum RMS diameter of the spectrometer system imaging slit is less than 17 mu m. The modulation transfer function (MTF) is greater than 0.99@3.7 lp/mm, and the maximum color distortion is 1.1 mu m. The energy concentration in three pixels is over 96%, which can be used for spaceborne high-resolution spectral detection of atmospheric components. The spectrometer system adopts a damping truss-unlocking mechanism for three-point support and is installed on the bottom plate of the satellite load compartment. The front lens tube is made of a carbon fiber composite material to ensure the thermal stability of the primary and secondary lens spacing. The main bearing frame and connecting plate are made of a titanium alloy. The design of stray light adopts the combination of "secondary mirror mask and primary mirror central hole baffle" with the simplest structure, which can ensure that the stray light coefficient of the camera is less than 0.5%. The statistics of various light paths that could reach the image surface were also obtained, and no abnormal stray light paths were found. Conclusions For space-based applications of the filament LiDAR spectrometer system, the optical system was designed and examined, and the main technical indicators of the optical system were determined. The designed spectrometer system can achieve a spectral resolution of 2 nm in the spectral range of 320-950 nm and thus provide a reference for the design and development of spectrometers used in filament LiDAR systems.

The response time of the electron to light in photoemission is difficult to define and measure. The tunneling ionization of atoms and molecules in a strong laser field is a type of strong field-induced photoelectric effect. In this process, the electron response time will change the time of high-order harmonic generation (HHG), which will have a fundamental influence on the reconstruction of electron attosecond dynamics through HHG. We propose a simple theory to resolve the response time problem in strong field atomic tunneling ionization. The response time corresponds to the strong interaction time of three bodies ie Coulomb, electron and laser field, which can be determined at the quantum-classical boundary. The observable directly obtained through response time can quantitatively reproduce a series of attoclock experimental curves and provide consistent explanations for these experimental phenomena. This work introduces the main conclusions of response time theory and summarizes in detail the research progress of this theory. Firstly, this theory can be applied to the orthogonal two-color laser field to quantitatively explain the main characteristic structures of photoelectron momentum distribution ( PMD). Besides, with this response time theory, the scaling law of the observable in attoclock experiment can be obtained. The proposal of scaling law is expected to provide a systematical theoretical guide for better understanding the applicability or feasibility of the attoclock under different conditions . In addition, based on the atomic response time theory, we further consider the property of multi-center Coulomb potential of molecular and develop a response time theory suitable for molecular system. Subsequently, we further apply the response time theory to polar molecules, by utilizing the asymmetry of PMD closely related to response time to recognize the permanent dipole (PD) effect within the laser sub-cycle. In the end, we discuss the prospects for research on response time. Firstly, it is envisioned to further apply response time theory to weak light and single photon transition to detect the response time of related processes. Besides, considering the significant influence of response time on the property of time-domain of HHG electron trajectories, the recombination (re-scattering) effect based on the current strong field tunneling ionization response time theory can be further investigated, thus extending this theory to describing HHG and above threshold ionization (ATI) processes. Furthermore, designing the "re-scattering electron trajectories" reconstruction scheme based on the electron trajectories with response time correction will provide important suggestions for HHG spectroscopic experiments. Finally, considering the asymmetric ionization caused by the PD effect of polar molecules, if the net ionization yield of adjacent sub-cycles is used as the current indicator,polar molecules can be used as a "micro diode" to study a type of attosecond response switching device. Polar molecular diodes emit electrons through tunneling ionization in laser field. According to the response time theory, tunneling occurs almost instantaneously, and response time needs considering only at the tunneling exit. Based on this, by searching for suitable materials (such as two-dimensional materials), it is possible to design a type of semi-classical diode (which can utilize tunneling) with femtosecond or even sub-femtosecond response time. The response time theory can provide a convenient theoretical tool for designing of such tunneling diodes.