This PDF file contains the front matter associated with SPIE Proceedings Volume 13254, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

Graphene oxide (GO) has great potential in organic optoelectronic devices as a hole injection material. However, it is difficult to achieve high quantum efficiency for organic light-emitting didoes(OLED) using GO alone because there is a big energy level mismatch and large roughness between GO and light emission layer. In this work, a composite hole transport material combining GO with water soluble copper phthalocyanine (TSCuPc) was adopted to achieve energy level matching between ITO anodes and emission layers, and further to realize both high hole injection efficiency and high hole transport efficiency. It is shown in our results that the effective HOMO energy level of GO varies with its atomic layer thickness, therefore good energy level matching between GO with ITO anodes and low energy barrier for holes injection can be achieved by adjusting GO layer thickness. Combining TSCuPc having appropriate thickness with GO can significantly reduce the roughness of hole injection interface and improve the directional migration rate of holes from the anode to the emitting layer, which thereby enabled the devices to achieve higher injection currents at the same operating voltage. Compared with the devices using GO alone or TSCuPc alone as the hole transport layer, the OLEDs with the composite hole layer of GO and TSCuPc obtained a higher current-luminance conversion efficiency and 2.2Voflow turn-on voltage, which indicates that the composite hole transport layer of GO and TSCuPc is effective to improve the quantum efficiency of OLEDs. The property of adjustable HOMO level of GO and the property of strong charge coupling between GO and organic molecules are beneficial to construct high efficient composite hole transport materials.

Real-time monitoring of downhole temperature and pressure variations plays a crucial role in ensuring well safety and effective reservoir management in oil and gas exploration. Leveraging the fiber Bragg grating technology with multilongitudinal mode laser frequency modulation and incorporating side-hole optical fiber grating sensors enable highresolution monitoring of well temperature and pressure. Utilizing digital demodulation technology to decipher multilongitudinal mode frequency modulation signals addresses challenges associated with traditional spectrum analyzers, such as structural complexity and poor stability. However, current research lacks dedicated demodulation methods for multi-longitudinal mode laser frequency modulation signals, apart from general software processing approaches. This paper introduces a hardware acceleration approach for frequency modulation demodulation. It outlines a hardware acceleration architecture based on a Field-Programmable Gate Array (FPGA). This architecture implements hardware circuits for various demodulation processes, including all-phase fast Fourier transform and phase difference method spectrum correction. It adopts a pipeline architecture and multi-clock domain processing to enhance signal processing efficiency. The system ultimately achieves a demodulation frequency of 15.259 kHz, a demodulation delay of 0.143ms, and a data utilization rate of 100%.

When the robot performs real-time pipeline weld tracking, the weld image obtained by the camera will be interfered with by solid arc noise and spatter, which makes it challenging to ensure the stability of the weld quality of the pipeline. This paper proposes an automatic weld seam feature recognition algorithm based on an improved U-Net neural network. The method extracts the global features of the weld image through the backbone network after down-sampling and upsampling in the U-Net network, fuses the laser stripe information at multiple scales, and utilizes the feature enhancement module to obtain more explicit weld feature images. Experiments have shown that the accuracy can reach 99.17%.

Laser diodes (LDs) can not only be used for solid-state illumination, but also have broad applications in visible light communication (VLC) due to large modulation bandwidth. In order to realize the bifunctional applications of both laser illumination and visible light communication, the red-green-blue (RGB) LDs of synthetic white light should have the same shape and uniform intensity in the illumination region. Here, we propose and demonstrate the beam shaping and uniformity function at red-green-blue wavelengths of 450 nm, 532 nm, and 625 nm by a single two-dimensional diffractive optical element (DOE). By using the iterative algorithm in conjunction with the optimization algorithm and variable weight, the proposed device can achieve an average energy efficiency of 81.27% and an average light spot uniformity of 91.64% under RGB trichromatic lighting conditions. Experimental fabrication of the proposed DOE was fabricated using UV laser direct writing photolithography, and the experimental results show that an average energy efficiency of 60.25% and an average uniformity of 83.76% of a rectangular spot at three wavelengths of RGB can be obtained, and excellent beam shaping and homogenization under composite white light illumination can also be achieved by the fabricated two-dimensional DOE.

A tunable wideband terahertz absorber based on vanadium dioxide (VO₂) has been proposed, and its optical characteristics have been studied. The absorber structure is comprised of three layers, which includes a VO₂ metasurface layer, a silicon dioxide (SiO₂) dielectric layer, and a gold substrate. The structure of this VO₂ metasurface consists of symmetrically distributed square perforated VO₂ resonators with circular holes, with a relative dielectric constant δ < 3.8 for the SiO₂ dielectric layer and a conductivity 4.56 10⁵ S/m 5 ≥ for the gold substrate. We studied the optical characteristics of the absorber through simulations and theoretical calculations employing the Finite-Difference Time-Domain (FDTD) technique. It was found that, while the VO2 is in the metallic phase, the absorber demonstrates a superior absorption spectrum, achieving a high absorptivity of 90% across a frequency range from 2.2 THz to 4.0 THz, with the absorption bandwidth reaching up to 1.8 THz. The absorptance for two peaks at 2.7 THz and 3.4 THz reaches 99% and 100%, respectively. When VO2 transitions to its dielectric state, its absorptance can be adjusted in a dynamic manner from 100% to 20%, achieving almost perfect amplitude modulation. The physical mechanism of wideband absorption is elucidated using electric field distribution and tunable absorption is verified using impedance matching theory. Besides, it displays insensitivity to variations in the angle of incident light polarization and maintains absorption stability under oblique incidence. This absorber exhibits excellent absorption performance, and the research and application of terahertz devices have opened up new frontiers.

The design requirement for the surface accuracy of optical components in the projection objective is to reach the nanometer level, which poses extremely high requirements for the surface accuracy of integrated optical components. In the projection objective system, the window lens group is a key component to ensure the normal operation of the optical mechanical system. It not only protects the internal environment of the objective but also participates in the system imaging. A flexible support structure for an integrated three-point positioning window lens is designed in the projection objective. The natural frequency of the overall structure and the influence of the supporting structure on the surface shape of the lens are analyzed. In optical system error analysis, the accuracy of surface shape prediction is achieved by analyzing the similarity between the upper and lower parts of the window lens. A multi-objective optimization algorithm has been used to improve the accuracy of lens surface shape and the stiffness of the lens group. Isodomain analysis is conducted on the support surface shapes of window lenses of different sizes, which can be served as theoretical basis and empirical data for the design of similar structures. The method proposed in this paper for designing, optimizing, and analyzing the support structure of window lens can provide a theoretical basis for the development of window lens support structures in projection objective.

DLC (diamond-like carbon) films have excellent optical properties, especially in the mid-infrared band (8-14 micron). Because of the adjustable refractive index in a certain range, it can be matched with more infrared materials, which is a hot research topic in infrared optics. In this paper, the effects of technological parameters and mixed gases (H2 and Ar) on the structure and properties of DLC films were studied. The results show that the process parameters can change the structure of the film, so that some properties of the film can be changed accordingly, but DLC films with high thermal stability and low internal stress can not be obtained simultaneously. When other gases are mixed in the reaction source gas, the structure and properties of the film will change, which is mainly due to the different role of the introduced gas in the film deposition process.

According to the requirement of high precision laser thickness gauge, based on the principle of laser triangulation ranging, design a double laser triangular thickness measurement optical system . The system uses two laser triangulation probes with different wavelengths to scan the upper and lower surfaces respectively. The thickness of the object to be measured is calculated by the size of the echo signal of the reflector. Using the optical design software Zemax to design its optical system. The zoom structure is introduced to replace the fixed focus system currently used in the thickness gauge on sale, so as to adapt to different measuring objects. The system has a focal length of 15-60 mm, no special glass and aspheric surface, which greatly reduces the processing and manufacturing costs. When Nyquist frequency is 60 lp/mm, the meridian modulation transfer function curve and sagittal modulation transfer function curve of each field of view are above 0.6, close to the diffraction limit, and the imaging quality is good. Verified by experiment, the measurement resolution is 0.1 micron and the measurement accuracy is 1 micron. It can be widely used in various thickness measurements in industrial testing.

In order to realize precise displacement measurement with low cost, a precise displacement sensor composed of right-angle reflectors is proposed in this paper. The sensor was designed as a reading head composed of a laser source and a position sensitive detector. The geometric optical amplification principle is proposed in this paper, when the relative displacement of the laser relative to the right angle mirror occurs, the relative displacement of the outgoing laser reflected by the right angle mirror will be multiplied. The displacement sensor with even magnification effect can be constructed only by the optical characteristics of right-angle mirror. The displacement measurement principle based on geometric optical amplification proposed in this paper, and the sensor prototype with double amplification is constructed, and relevant experiments are carried out. The experimental results showed that the sensor could achieve twofold amplification, the displacement measurement accuracy was ±500 nm, and measurement resolution was better than 100 nm. High-precision displacement measurement and a wide range can be achieved by using multiple reading heads. The measurement system does not require high precision of optical devices, so it is especially suitable for constructing a displacement measurement system with small range and high precision, through the multiple amplification structure, more than 10 times geometric optical amplification can be realized.

Because of the sparse distribution and lack of organization in point clouds, 3D object detection has been a challenging task. Current detection models heavily rely on 3D labeled data, while in real-world complex scenarios, issues such as point cloud sparsity and occlusions make feature extraction difficult. In this study, we propose a pre-training approach based on self-supervised masking. We compute the distance between voxels and the LiDAR device to regulate the masking ratio of voxels, randomly masking voxels accordingly. Predicting the occupancy status of voxels enables the prediction of the masked occupancy structure for the entire 3D scene. This masking strategy enables self-supervised model training, achieving the effect of data augmentation. We conducted thorough experiments using the KITTI datasets, specifically targeting the 3D object detection task. We achieved a significant +2.11% improvement in the Hard category of Car detection. mAP across three difficulty levels also increased by +0.74%, illustrating the efficiency of our method in handling complex circumstances.

This paper, a high-energy particle detection optical system is designed to solve the high-energy ion beam in the nuclear reactor magnetic field and bombardment of some regions of ZnS scintillator, the region will emit fluorescence, through fluorescence detection and imaging, so as to analyze the relevant physical characteristics of high-energy particles. The high-energy particle detection optical system includes object surface, imaging group, relay mirror group, protection mirror group, correction mirror group, and detector assembly, spectral range of 400 nm to 550 nm, field angle of 40 degrees, pupil diameter of 10mm, system F number of 4, and total system length of 2085mm. Except for the detector assembly, 13 lenses and 1 protective window were used. The high-energy particle detection optical system designed in this paper has the advantages of easy processing, easy loading and tuning, small lens aperture, and good imaging quality.

In this paper, the optical system of an unmanned airborne greenhouse gas hyper-spectral imager that can be used for greenhouse gases is presented. The spectrometer adopts a self-collimating Offner structure. The telescope system adopts the image square telecentre structure. The spectral resolution is improved to more than 0.2 nm by image merging, and the signal-to-noise ratio is better than 200 at an integration time of 1 s at the simulated incident pupil brightness.

Laser decontamination technology is used to remove the oxidized layer on the surface of stainless steel, and the influence of different combinations of process parameters on the decontamination effect is investigated. The experiment adopts orthogonal experimental design, through the scanning electron microscope and energy spectrum analyzer to analyze the surface morphology and composition distribution of the specimen surface before and after laser decontamination; the use of metallurgical microscope to observe the cross-sectional organization of the specimen before and after laser decontamination; the use of BRUKER white-light interferometer to decontaminate the specimen three-dimensional morphology analysis, test its decontamination thickness and surface roughness. The results show that under the optimal process parameters (laser power of 180W, repetition frequency of 500kHz, travel speed of 5mm-s^-1, pulse width of 300ns), the oxidized layer on the surface of stainless steel is completely removed, and the thickness of the decontamination reaches 4.1 μm, and the surface roughness is reduced to 0.28 μm, and the surface of the decontaminated surface mainly consists of Fe, Cr, Ni and other matrix elements. Laser decontamination for stainless steel surface oxides have excellent decontamination effect.

Lithium-ion batteries are extensively utilized for their advantages such as high energy density, environmental protection and long cycle life. there are pressing issues that need immediate resolution such as safety, stability and reliability of lithium-ion batteries. In view of this, this paper studies a fiber Bragg grating (FBG) sensor implanted in 280 lithium-ion battery to measure cell strain technology, through the reference grating based method to overcome FBG temperature and strain cross-sensitivity problem, this method can be used to analyze the operating characteristics of lithium-ion battery.

In our research, we synthesized tris-(8-hydroxyquinoline) lanthanide (Lnq) (Ln = Eu, Gd, Tb).The electron acceptor and donor of an organic ultraviolet (UV) photovoltaic diode (PV) are Lnq and 1,3,5-tris(3-methylphenyl-phenylamino)- triphenyamine (m-MTDATA).We compared the photocurrent of the device to the PL spectrum. The best PV comes from m-MTDATA: Gdq mix, which has a stable 4f electron structure and lower monomer or exciplex emission.

In spectral domain optical coherence tomography (SD-OCT), spectrometers traditionally use gratings to disperse light, producing spectral signals linearly related to wavelength. However, this induces nonlinearity in wavenumber (k) space, diminishing the sensitivity of SD-OCT systems. Conventionally, signals are interpolated and resampled to achieve linear interferometric spectral signals in wavenumber space. Our work presents a spectrometer that inherently produces linear wavenumber signals, obviating the need for post-interpolation and thus enhancing system sensitivity while reducing processing time. This advancement is achieved by employing a spectrometer equipped with a freeform surface lens that dramatically reduces wavenumber nonlinearity from 1.49% to 0.0043%. We have successfully manufactured this lens and experimentally validated the spectrometer’s wavenumber linearity.

In perovskite light-emitting diodes (PeLEDs), a rational device architecture design is fundamental to the fabrication of high-performance devices. Recently, Self-assembled monolayers (SAMs) have been utilized in PeLEDs to enhance device performance due to their ability to modulate energy level alignment and their spontaneous formation into dense films. However, during the solution processing, SAMs because of their self-assembly properties, tend to form micelles, which can compromise the integrity of formed SAM. In this study, it has been observed that the introduction of methyl groups into the molecular structure of phosphonic acid-based carbazole SAMs helps to disrupt micelle formation, thereby leading to the creation of dense SAM. The use of SAM also strengthens the robustness of the poly(9-vinylcarbazole) layer, subsequently reducing the defect density in the perovskite layer and suppressing the process of non-radiative recombination. Consequently, PeLED devices developed with SAMs have demonstrated improved performance. These discoveries make a contribution to the development of new SAM materials for use in PeLEDs.

In the construction of multi-target tracking system, the Particle Filter-Probability Hypothesis Density (PF-PHD) filter is difficult to meet the real-time requirements of the system due to its large number of particles and complex filtering process. In this paper, we propose a real-time realization technique of PF-PHD based on computational structure analysis, which firstly improves the particle newborn process in the prediction step of PF-PHD, and proposes an observationdriven particle newborn method based on the pre-newborn particle template. Subsequently, the computational structure analysis of PF-PHD is carried out, and the hardware acceleration scheme based on FPGA is designed according to the data flow and control flow in the filtering process to complete the hardware and software division steps. Finally, the system realization scheme is compared with DSP respectively, when the FPGA running frequency is 150 MHz and the DSP running frequency is 1 GHz, the speed of the hardware acceleration module is about 6.81 times of the DSP processing speed, and the system realization speed is 1.66 times of the DSP realization speed.

Topological edge states (TESs) have garnered significant attention for their robust and unidirectional transport properties, often serving as ideal waveguides to mitigate the backscattering caused by local structural disorders in recent years. However, the influence of waveguide dispersion and edge bending on the pulsed TES has not been discussed yet. This paper investigates the above-mentioned issues using the classical Kane-Mele model and Gaussian pulses as examples. On one hand, the nonlinear dispersion of TESs not only results in an increase in pulse width but also leads to the deformation of Gaussian pulses. On the other hand, edge bending at various angles can either amplify or diminish the pulse width. This study offers two effective methods for adjusting the pulse width, providing new perspectives on manipulating and controlling light propagation in topological waveguides.

Metasurfaces including chiral metasurfaces have attracted extensive attention due to their ability to flexibly manipulate various parameters of light fields, such as amplitude, phase and polarization, based on the subwavelength-scale structures. However, conventional design and optimization of a 2D metasurface usually rely on experience-driven and time-consuming trial-and-error procedure for searching effective structures and the followed case-by-case optimization, in which the structures are usually limited to simple and regular geometries. Here, we propose a design and optimization method of chiral metasurfaces with the combination of structural discretization and genetic algorithms. Based on the proposed method, design and optimization of a broadband chiral metasurface working in the visible wavelength region was conducted with an arbitrarily given C-2 symmetric structure without complicated structural searching and time-consuming optimization process. Simulation results show that an average circular dichroism of 0.73 (peaking at 0.88) and an average circular polarization extinction ratio (CPER) of 11.5 dB (up to 27.4 dB) can be obtained in the wavelength range spans from 605 nm to 720 nm. This proposed method provides a new approach for the design and optimization of chiral metasurfaces or other metasurfaces.

Osteoarthritis (OA) is a common chronic degenerative disease worldwide. Currently, traditional diagnostic methods for OA have disadvantages such as high cost and long time consumption. For this reason, it is urgent to have a method that can diagnose osteoarthritis quickly and accurately. During this study, we suggest a multi-column convolutional neural networks (MCNN) model based on Mel-frequency cepstral coefficient (MFCC). We first compare the effectiveness of Partial Least Squares (PLS) and MFCC in preprocessing the data and comprehensively evaluate the performance of the model by five-fold cross-validation. To further improve the ability of recognizing spectral features, We utilize the Spectral Segmentation Method (SSM) to construct the MFCC feature sequences into 2D spectral feature maps and input them into the MCNN for classification. The final experimental results show that our proposed mfc - ssm - mcnn model has a high classification efficiency in diagnosing osteoarthritis with an accuracy of 95.0%.

Raman spectroscopy is a widely used analytical technique that provides extensive information about the chemical composition and molecular structure of samples. It is based on the Raman scattering phenomenon, where when a sample is irradiated with excitation light, photon scattering occurs, causing a slight shift in frequency that reflects the vibrational and rotational states of molecules in the sample. By analyzing these frequency shifts, one can understand the types of chemical bonds, molecular configurations, and other relevant information within the sample. Osteoarthritis, the most prevalent joint disorder, is typified by the degradation of articular cartilage and the engagement of all tissues within the joint. eventually leading to cartilage degeneration, fibrosis, rupture, defects, and damage to the entire joint surface. Therefore, timely and accurate diagnosis and treatment of patients are crucial. In this study, we aimed to achieve an objective, rapid, and accurate diagnosis of osteoarthritis using serum Raman spectrum combined with deep learning methods. In this experiment, serum samples were collected from 116 osteoarthritis patients and 116 healthy control subjects, and Raman spectroscopy data were obtained. The collected spectral data were preprocessed by the adaptive iteratively reweighted penalized least squares (airPLS) and Savitzky-Golay (SG) filtering algorithms. CNN and TCN classification models were selected to classify and identify osteoarthritis patients and healthy controls. The results showed that TCN had the excellent identification performance, with an average accuracy, sensitivity, and specificity of 97.03%, 100%, and 82.86%, respectively, over five experiments. The area under the ROC curve (AUC) was also the highest at 0.97. These experimental results indicate that deep learning methods based on serum Raman spectroscopy have great potential in the rapid diagnosis of osteoarthritis and can provide reference for the auxiliary diagnosis of other diseases.

In order to ensure the safety of the road surface, it is very important to detect the cracks in time and accurately. This paper proposes a pavement crack detection network model based on multi-scale feature extraction and deep supervised feature fusion. Firstly, multi-scale separable convolution blocks are used to extract crack features, which makes the model more effectively model pavement cracks with different topological structures. Then, the obtained feature maps are added to the deep supervised network to aggregate multi-level features, so that the model can converge faster and better. Finally, in order to make full use of the lateral output feature information of each stage, the dynamic feature fusion method is used to realize the efficient integration of multi-level features, which significantly optimizes the completeness and accuracy of the final prediction map. According to the evaluation results on DeepCrack, CFD and Crack500 public datasets, the proposed method shows better performance than other methods in crack detection.

Accurate detection and tracking of moving targets in dynamic environments are essential for various applications, including military surveillance, autonomous navigation, and robotics. In order to address the contra-maneuvering target tracking challenge in 120mm rocket guidance systems based on image strap-down guidance, which aims to identify and track maneuvering targets in TV images and provide bias information for back-end flight controllers, this paper proposes a threshold for binary segmentation based on images. The target is segmented within the tracking box using binaryvalued images. Additionally, an adaptive weight Kalman filter algorithm for target tracking is proposed. By dynamically adjusting filter weights based on target motion characteristics, the proposed approach effectively addresses challenges such as occlusion and rapid maneuvering, ensuring stable and reliable tracking throughout the trajectory. Comprehensive simulations demonstrate the efficiency of the proposed algorithm, illustrating its ability to accurately track maneuvering targets under varying conditions. The results highlight the superiority of the proposed approach in providing precise target localization and trajectory prediction, thus offering valuable advancements in image-based guidance systems.

Optical vortex beam with orbital angular momentum (OAM) has great potential to increase the capacity of optical communication and information processing in classical and quantum regimes. Nevertheless, important challenges that influence the optical data transmission in free space is the existence of diffusers along the optical path, which causes inevitable information loss during the wave propagation. Numerous algorithms have been proposed successively for identifying the modes of vortex beams propagating through scattering media. However, these methods all require completion on a computer, which is energy-intensive and energy consuming. Here, we propose an all-optical regime for identifying the modes of vortex light fields propagating through scattering media. After training by deep learning, our model can recognize the mode of vortex beam through unknown phase diffusers, demonstrating generalization to new random diffusers that have never been encountered before. Once physically deployed, the entire setup will rapidly identify the modes of vortex light propagating through scattering media at the speed of light, and the entire inference process will consume zero energy except for illumination source. Our research represents a significant step towards highly accurate recognition of vortex light modes propagating through complex scattering media, providing significant guidance for the application of optical communication in complex environments.

Hyperspectral imaging is an efficient way to overcome the limitations of detecting different object with similar visible light texture. The study aims to expand the feasibility of hyperspectral imaging to classifying the stains and defects on mobile phone cover glass. Firstly, we extracted eight optimal spectral features by decision tree method, including 526 nm, 567 nm, 582 nm, 629 nm, 689 nm, 711 nm, 789 nm, and 888 nm. Our classification used the Random Forest modeling method (RF). Experimental results showed that, based on optimal spectral features, the precision of RF model outperformed for classifying stains and defects (Precision > 0.9). Overall, this study contributes a reliable and convenient tool for classification of stains and defects on mobile phone cover glass, offering scientific insights to support quality control inspection for mobile phone cover glass.

In the detection of the temperature rise of the reflective element, due to the lack of effective numerical simulation methods, the test is carried out by the measured method, resulting in a large loss of time and economy. Therefore, an accurate and effective numerical analysis method is needed to estimate the temperature change of the element under laser irradiation. This paper firstly conducted a temperature rise experiment on a reflection mirror plate of a specific film system under 1030 nm wavelength laser irradiation; secondly, the reflection mirror was modeled and numerical calculations were analyzed to determine the size of the boundary conditions for numerical analysis; finally, the effects of film system structure, air convection coefficient, and convection mode on the temperature rise of the mirror plate under laser irradiation were explored. The results show that the method proposed in this paper can accurately simulate the temperature change of specific film mirrors under continuous laser irradiation. The film structure has a great influence on the temperature, and the error is 47.4 %. The air convection coefficient and different pairs of manifolds have a great influence on the temperature change during long-term irradiation, and have little effect on the temperature in a short time, which can be ignored.

Vegetation is the core component of the surface ecosystem, which plays a very important role in maintaining the stability of the ecosystem. With the increasing social and economic activities of human beings, scientific, accurate and efficient detection of vegetation ecology mutation has become particularly critical. Using Terra satellite MOD13Q1 data, a method for detecting vegetation ecology mutation based on NDVI (Normalized Vegetation Index) long and short time series data is established. It can not only analyze the vegetation change law and characteristics by using NDVI long time series data, but also detect whether there is mutation in NDVI short time series data, and determine the mutation time, position and ground object type before and after mutation. The experimental results of vegetation ecology mutation in Block 1 from 2005 to 2007 and Block 2 from 2015 to 2017 in typical areas of Hetian, Xinjiang show that the vegetation ecology of Block 1 changed suddenly in 2006 compared with that of 2005, and the vegetation type was woodland, but there was no mutation in vegetation ecology in 2007 compared with that of 2006, and the vegetation type did not change. Compared with 2015, the vegetation ecology of Block 2 changed suddenly in 2016, and no matching vegetation type was found. Compared with 2016, the vegetation ecology changed suddenly in 2017, and it was judged as wasteland. The results are consistent with those of high-definition satellite images and field investigation. This method can detect the vegetation ecology mutation in the target area scientifically, accurately and efficiently, which is of great significance to the protection and rational development and utilization of ecology resources in the process of regional economic construction.

In the field of non-contact 3D reconstruction, structured light 3D reconstruction combining phase-shifting and Gray code is a high-precision reconstruction method. To address the problem of phase jumps easily occurring during phase unwrapping with traditional Gray codes, this paper proposes a cyclic complementary Gray code phase unwrapping algorithm based on gradient correction. Firstly, the wrapped phase image is converted based on the cyclic complementary Gray code to obtain the absolute phase image. For fixed-value phase jumps caused by blurred fringe order edges, phase jump points and their directions are determined through the second-order gradient of absolute phase, followed by phase correction to obtain smooth and continuous absolute phases. In the reconstructed 3D point cloud of the gypsum sculpture, the number of discrete points decreased by 96%. The experimental results indicate that the algorithm effectively discriminates phase jump points and corrects phase jumps without relying on pre-calibrated threshold values.

Recent research in all-dielectric asymmetric metasurfaces has demonstrated the capability to generate highly sharp Fano resonances, offering bright prospects for applications in optical biosensing. This work proposed a Fano Resonance in Near-Infrared Metasurface based on Asymmetric All-Dielectric Cylindroids. Each unit of the metasurface consists of two all-dielectric Si elliptical cylinders with different short-axis lengths arranged on top of an MgF₂ dielectric layer. By employing the Finite-Difference Time-Domain (FDTD) numerical analysis method, we investigate the optical characteristics of the metasurface. We found that when the semi-minor axis of the asymmetric cylindroids are 0.1μm(w₁) and 0.094μm(w₂), the metasurface exhibits a sharply narrow Fano resonance peak at λ=1.013μm, with a reflection intensity exceeding 92% and a Q-factor as high as 580. Which works at near-infrared region. The physical mechanism of the metasurface is the principle of electromagnetic coupling. The simulation results indicate that the Fano resonance arises from the interference of two distinct electric quadrupole modes. Moreover, the results demonstrate that the sensor exhibits a sensitivity of up to 85 nm/RIU, thereby validating its potential applications in areas such as biosensing and refractive index sensing.

The application of visible light remote sensing technology in unmanned aerial vehicles (UAVs) has become extensively utilized across a variety of sectors. However, the research on its development and hotspot is very limited. Based on the literature of visible remote sensing of UAVs collected in Web of Science (WOS) database from 1998 to 2022, the literature knowledge map was drawn by visualization analysis software CiteSpace to reveal the research progress and trend in this field. The results show that: (1) The research has cross-application in multiple fields and multi-source data fusion, which main focus is on expanding research perspectives, objects and methods; (2) Research topics mainly focus on monitoring, photogrammetry and agriculture; (3) UAV image, model building and application, vegetation index, classification, etc., are research hotspots, and the main research directions in the future are machine learning, point cloud, virtual reality, etc. The results can provide reference for researchers to carry out further research in this field.

Skeleton-based action recognition utilizes human skeletal data to identify and analyze human behaviors , which is widely applied in various scenarios such as intelligent surveillance, virtual reality, and autonomous driving. Since human actions usually have different durations, recognizing skeleton-based actions accurately and efficiently remains a challenge. To address the issue of temporal scale, this paper introduces a temporal multi-scale feature network (TMSF-Net) designed to enhance the recognition of skeleton-based actions. Specifically, TMSF-Net introduces a multi-scale temporal convolution module (MSTCM) to flexibly adjust the temporal receptive field of the network, enabling it to focus more on action-related regions. Additionally, TMSF-Net incorporates a Global Filter Module (GFM) to enhance the interaction among joint points across spatial and temporal dimensions. and adapt to different action scenarios. The efficacy of the proposed approach is demonstrated through experimental validation on two public datasets dedicated to skeleton action recognition.

Self-supervised monocular depth estimation uses only one camera to get depth information. It does not need manually real depth information as training data, but is trained by the geometric information contained in the image itself. While many existing methods use heavy backbone networks for precision, designing lightweight models can reduce the computational and memory consumption, making them suitable for resource-constrained environments or embedded devices. In this work, we propose a lightweight network (FasterDepth) for self-supervised monocular depth estimation. Additionally, in order to merge the rich information of multi-stage of the network, this work raises a multi-stage feature fusion module. Experiments on the KITTI dataset show that our FasterDepth has higher precision and fewer parameters than Monodepth2.

The preparation technology of high groove density gratings is generally complex and costly, while nanoimprint technology has the ability to prepare large quantities of high-precision micro-nano structures. Therefore, nanoimprint technology is used to study the replication process of grazing incidence high groove density gratings used in Littman-type external cavity semiconductor lasers (ECDL). A UV composite rolling technology is used to imprint the high groove density grating groove area on the ULE glass substrate, and the grating replication is completed after coating. Replicated grating performance was tested and characterized with high fidelity.

The purpose of scene flow estimation is to capture the intricate motion patterns within point clouds across successive frames. We incorporate the channel self-attention (CSA) into the estimation of scene flow for point clouds. Specifically, the channel self-attention mechanism prioritizes channels with significant disparities to prevent the merging of similar and redundant information. Through the subtraction operation embedded in the structure, attention weights are concentrated in regions with salient characteristics and crucial information within the point cloud, thereby reducing attention toward noise points. By incorporating channel self-attention at each stage of the network, we can extract local features and capture rich contextual information. Additionally, we introduce a channel excitation module to enhance the global correlation among channels and enhance the model's representation capability by introducing additional nonlinear relationships. Experimental results verify that our proposed method is effective.

In this paper, we designed a non-contact measurement system of alcohol concentration based on near-infrared absorption. The system measures the absorption spectra of alcohol solutions with known concentration, and explores the relationship between solution concentration and absorbance. The results show that the linear correlation between the absorbance value and the solution concentration is more than 0.99, which is in good agreement with the theory. The least-squares method was used to fit the absorbance and alcohol concentration to establish the functional relationship. Besides, the incident light intensity and transmitted light intensity of the alcohol solution are measured to calculate the absorbance of alcohol solutions of different concentrations and then obtain solubility information of alcohol solutions. This system is characterized by simple structure and accurate and reliable measurement, which can effectively improve the efficiency of solution concentration measurement. In addition to measuring the concentration of alcohol solution, this program can also be applied to measure the concentration of other solutions, only with a light source of appropriate wavelength and the corresponding photoelectric detector, so the design of the program has a very broad application prospects.

In multi-object tracking, the one-shot method combines detection and appearance embedding, simultaneously locating the target and extracting appearance features in a network, which improves the efficiency of the multi-object tracking algorithm. However, the previous method did not fully utilize the detection information. We propose multi-level matching mechanism is proposed to improve tracking performance by processing targets hierarchically. The results on the MOT17 dataset show that our method achieves competitive results.

Focused on the crack defects commonly encountered during the manufacturing process of laser additive manufacturing (LAM), this paper introduces a laser ultrasonic imaging technique specifically tailored for these defects. This method relies on cross-correlation analysis for precise detection. Utilizing the selective laser melting (SLM) technology as a case study for printing AlSi10Mg alloy, the surface of the specimen was meticulously scanned using laser ultrasound, allowing us to capture the ultrasonic surface wave signal. Subsequently, the crack defect imaging is achieved through a cross-correlation analysis of adjacent surface waveforms. To address the challenges posed by background noise and sample surface roughness, an adaptive threshold method is introduced. This innovative approach effectively reduces noise in the defect imaging results, thereby significantly enhancing the imaging accuracy of crack defects. The findings reveal that the cross-correlation and adaptive threshold-based defect imaging method not only sharpens the boundary features of defects but also significantly diminishes the shadow area during crack defect detection. This offers a more efficient and convenient approach for defect detection and recognition, paving the way for improved quality control in LAM processes.

Chalcogenide glasses exhibit high optical nonlinearity and are widely used in developing various optical modulating devices. Due to the influence of coating processes and the properties of the glass material, the film thickness is one of the factors affecting the optical nonlinear coefficient of chalcogenide glass thin films. This paper presents the nonlinear optical characterization of chalcogenide glass Ge₁₂Sb₂₈Se₆₀ thin films of various thicknesses. The measurements were conducted using the Z-scan method and the pump-probe method at a wavelength of 532 nm. The results demonstrated that as the film thickness increased from 25 nm to 200 nm, the nonlinear absorption of the material decreased from 1.1 × 10^-6 m/W to 0.68 × 10^-6 m/W. This trend will become more gradual as the film thickness increased. This means that the chalcogenide glass on the surface of the film has greater optical nonlinearity.

A novel approach utilizing a Gaussian filtering is presented to eliminate aberrations in the phase recovery for off-axis digital holography. After applying the Fourier-transform-based method to the off-axis hologram, a complex amplitude map containing object phase and phase aberrations can be obtained. A Gaussian filter is then employed to extract the phase aberrations map from the complex amplitude map. In this case, the object phase can be reconstructed through a division between the complex amplitude map and the phase aberration map. The proposed method does not require the sample-free hologram containing aberration information and avoids complex procedures such as surface fitting or spectral filtering. Experimental result shows that the proposed method can realize accurate and fast off-axis digital holographic phase compensation recovery.

In complex underwater environments, it is difficult for traditional methods to accurately obtain position information of dense, fuzzy, and small-sized organisms. Although the convolutional neural network algorithm is popular, it is limited by insufficient training samples and other limitations, and the accuracy and speed improvement are poor. For this reason, this paper designs a YOLOv7-CBF network model based on the YOLOv7 network. By introducing the CBIF module and FasterNet module, the model fuses local and global information, improves the feature extraction capability, and reduces redundant computation to effectively extract contextual semantic information. Meanwhile, a new enhanced loss function ECLOU is proposed to improve the localisation accuracy and model robustness. Experiments prove that the model performs well in underwater seafood detection with high accuracy and speed, which meets the practical needs. This result is of great significance for facilitating seafood fishing, reducing cost and improving detection efficiency.

Inorganic perovskite CsPbBr3 has excellent optical properties such as high photoluminescence quantum yield (PLQY), strong light absorption, adjustable band gap, narrow band emission, etc., and has wide application prospects in light emitting diodes, photodetectors, lasers and solar cells. In this paper, CsPbBr₃ nanocrystals (NCs) was synthesized by ligand-assisted reprecipitation (LARP) method, and the pump fluence and temperature-dependent photoluminescence (PL) properties of CsPbBr₃ NCs were studied by steady-state PL combined with spectral analysis and the existence of single-photon absorption was proved. The PL intensity decreases with the increase of temperature, the peak position of PL blue shifts with the increase of temperature, and the larger exciton binding energy is conducive to promoting radiation recombination. The longitudinal optical phonon energy of CsPbBr3 NCs is 28.8 meV, and the material has good thermal stability.

In order to improve the accuracy of recognition and segmentation of occluded and stuck cherry tomatoes in three-dimensional infrared images, an improved watershed segmentation method based on point cloud spatial density cluster analysis is proposed. First, a binocular infrared camera is used to collect a three-dimensional point cloud image, and then the point cloud holes are filled with geometric diffusion. The completed point cloud model is passed through a color region growing segmentation algorithm to extract the fruit bunch target. The phenotypic characteristics of the target are analyzed through point cloud density clustering, and then the concave-convex relationship is used to enhance the edge characteristics of the fruit. Finally, secondary clustering is performed based on the watershed idea to separate the blocked and adherent fruits. Experimental results show that the algorithm has a recall rate of 88.9% in target recognition of stuck cherry tomatoes, which improves the three-dimensional segmentation performance of stuck cherry tomatoes.

In the wave theory of light, the experiment of fiber optic Mach-Zehnder interference is an important experiment, which has the characteristics of complex experimental system and difficult interference adjustment. Based on Mach-Zehnder interferometer, this paper systematically studies the design and simulation of Mach-Zehnder experiment system for 650nm wavelength semiconductor laser source, and analyzes the coupling characteristics of semiconductor laser and single-mode fiber by using ZEMAX software, and tests the optical field distribution during the experiment. At the same time, the experiment of Mach-Zehnder interference of single mode fiber is realized.

The research findings regarding spatially suspended displays utilizing Er3+-doped microcrystalline glass are outlined in this article. Er3+-doped microcrystalline glass is employed as the substrate for spatial three-dimensional suspended displays, with fluorescence spectroscopy measurements conducted under the co-excitation of 1530nm and 852nm near-infrared light. Furthermore, an experimental setup for spatial three-dimensional suspended displays was devised to facilitate spatial imaging. Subsequently, a spatially suspended display with 360° viewing capability was achieved within the Er3+-doped microcrystalline glass suspension medium.

Aiming at the autofocus requirements of lenses with variable aperture and focal length, this paper proposes an autofocus method based on depth-of-field estimation to adaptively calculate the search step length and adjust the focusing region. First, based on the relationship between the focus curve broadening and the existence of depth-of-focus, the search step size during coarse focusing is adaptively calculated by depth-of-focus estimation, so that the motor quickly approaches the focusing peak. Second, using the clarity data obtained during the adaptive step-size search, the direction of object space extension can be estimated, so that focusing region optimization can be performed to improve the peak shift of the focus curve in the state of small depth-of-field. Finally, the hill-climbing search method with small step length is used to realize accurate focusing. Experiments show that compared with the traditional hill-climbing search methods and the coarse-to-fine search methods, this method can reduce the number of focus searches by about 35% through adaptive focusing method, thus speeding up the focusing speed, and can improve the focusing reliability in the case of small depth-of-field through region optimization.

Specially designed frame glasses have shown increasingly significant performance in myopia prevention and control in clinical trials. However, public studies on the modulation of high-order aberrations related to eyeglass frames remain scarce. This article designs eyeglass lenses with high-order aberrations and myopic defocusing by linking the eye model and frame glasses, and simulates the optical model of the glasses-eye for 300 degree myopic patients. When the highorder aberration modulation unit is not set for the glasses, the defocusing value of the Y-axis direction at a -14° field of view corresponding to the change in the external surface power of the lens defocusing unit microlens under static field, and the defocusing change of the external surface power of the defocusing unit micro lens is set to 6 diopters, with a static vertical field of view ranging from -28.5° ~ 28.5° (at 1° increments). This article discussed the correlation between the base arc curvature radius of the high-order aberration modulation unit’s toric microstructure and high-order aberrations in the designed eyeglass under static observation in the Y-direction. Corresponding empirical formulas have been established. This research will be conducive to the development of high-order aberration modulation glasses.

The existing optical cryptosystems always process the plaintexts into a random noise image, and the length of the key in the encryption progress is generally much longer than that of the plaintext. In this work, a Harr-wavelet-based optical cryptosystem using the convex lens is proposed. This scheme only needs recording the amplitude. The encryption key includes a series of parameters which is quite shorter than the length of the plaintext. These keys are natural inherent parameters. The Harr transform is introduced to improve the sensibilities of encryption keys and reduce the length of the ciphertexts. The ciphertext is a diffraction-pattern-like image which contains no information of plaintext. Besides, the cost of the whole optical encryption is reduced heavily. Extensive numerical simulations are carried out to verify the security, feasibility, and sensibilities of encryption keys of the proposed encryption scheme. In addition, how to properly select encryption keys is also discussed.

Liquid crystals (LC) planar optical elements have attracted much interests owning to its simple fabrication and powerful beam shaping capability. Among them, cholesteric liquid crystals (CLC) have received widespread attention due to their self-assembly behavior, natural chiral structure, circular polarization selectivity, and Pancharatnam–Berry (PB) phase modulation capability. However, traditional CLC lenses suffer from significant chromatic aberration which results in single wavelength or narrow bandwidth imaging only. Conventional way to solve the chromatic aberration is either combining LC lens with a refractive lens with opposite dispersion or designing multiple layers for a few discrete wavelengths, resulting in the optical system bulky and complicated. Here, we propose and demonstrate a single-chip achromatic CLC lens for broad and continuous wavelength band imaging. The proposed CLC lens utilizes cubic wavefront encoding technology and is prepared using optical orientation technology, avoiding the complex manufacturing process of conventional metalenses. Experimental results demonstrate that an F/#30 lens with a coding parameter of 20 can achieve achromatic imaging with a total bandwidth of 80 nm across the entire Bragg reflection band. This study opens an avenue for the achromatic manipulation of CLC lenses and has great potential in the field of imaging, display and dispersion management.

Research goal was to demonstrate whether multiphoton microscopy (MPM) can be employed for identifying T staging of esophageal cancer (EC). For comprehending T staging of EC better, the tumor infiltration depth in esophageal wall of each cancer staging was first shown. In order to automatic identification the boundaries between normal layers that without tumor infiltration, a boundary detection algorithm is proposed by extracting SHG signals which are produced by collagen molecules with asymmetric architecture in MPM images. Then, the morphological differentiation of tumor cells from main components of esophageal wall was performed. The tumor infiltration depth was decided by evaluating the position of tumor cells and the boundaries. After then, Tis, T1a, T1b, T2 and T3 staging of esophageal cancer were identified. The outcomes indicate that MPM can be applied for automatic identifying T staging of EC. Every effort was made to developed compact multiphoton microendoscope system with high speed and ultra-deep penetration depth in recent years. There was every reason to believe that clinical application of MPM in automatic identifying esophageal cancer staging will be possible in the future.

The vector 3D Gerchberg-Saxton (G-S) phase retrieval algorithm utilizes the properties of three-dimensional Fourier transform to perform more accurate calculations on the tiny light field, but this algorithm has low light intensity utilization and serious crosstalk in multi-plane holographic display. This paper proposes a modified vector 3D G-S algorithm, which improves image clarity by jointly adjusting the intensity of the reconstructed light field, and sets dynamic weights on each target plane to improve the image uniformity. In order to verify effectiveness of the modified algorithm, simulation comparison experiments are conducted on binary images and grayscale images in the case of 2-4 target imaging planes. The results show that the modified algorithm improves the clarity and uniformity of multi-plane image without slowing down the convergence rate, which is helpful for applications such as micro-nano processing that require high image quality.

An adaptive Fourier single-pixel imaging scheme based on pre-projection (ppFSI) is presented in this paper. This scheme preliminarily localizes the imaging region by pre-projecting a nine-grid pattern to the scene, then the target region is obtained by using a down-sampling method for the preliminarily localized region, and finally a high-quality reconstruction of the target region is obtained by secondary sampling of the target region. Compared with the existing single-pixel imaging methods, ppFSI can achieve high-quality reconstruction of the target region while significantly reducing the projected patterns because it only projects illumination patterns on the target region. The experiments strongly demonstrate the effectiveness of this scheme.

To improve the recognition accuracy of the embedded visual module and make it competent for visual tasks on complex occasions, an image preprocessing method used on the OpenMV is proposed. Aiming at the two main recognition tasks of machine vision, color recognition and shape recognition, an image preprocessing algorithm and a filtering algorithm based on the OpenMV of embedded vision module are proposed for use by analyzing the interference existing in practical problems. For color features, the processing of binary segmentation and image morphology can filter the background and the noise similar to the target, which greatly improves the recognition accuracy with only a small increase in recognition time. For shape features, kernel filtering and Hough Transform are used to filter the irrelevant targets in the image, which improves the recognition accuracy, reduces the image complexity, and speeds up the operation speed of Hough Transform. The experimental and practical results show that the image preprocessing method combined with OpenMV can not only take into account the advantages of the embedded visual module but also complete the task of target recognition with high accuracy, which helps meet the needs of practical engineering and promote the wide application of embedded vision.

Parallax processing and structure preservation pose significant challenges in image stitching. This paper introduces an image stitching method that addresses local alignment and line structure preservation in four-lens panoramic cameras through content-preserving warpping and the geodesic constraint. Recognizing that lines in fisheye images are curved(the curved lines are defined as the geodesic lines) and correspond to great spherical circles centered at the camera’s optical center, we utilize the Hough method to detect geodesic lines in the image warped by spherical projection. Subsequently, we incorporate the geodesic structure constraint into content-preserving warping to ensure the detected geodesic lines falling onto a large circle of the projection sphere and maintain line structures in panoramas during local alignment. Our experiments demonstrate that this approach effectively detects more and longer lines in images warped by spherical projection. Moreover, incorporating the geodesic constraint into content-preserving warping not only aligns images but also maintains local boundaries, ensuring that the lines in the perspective images exported from the panorama are straight.

Experimental Advanced Superconducting Tokamak (EAST) employs Extreme Ultraviolet (EUV) spectroscopic diagnostic systems to measure the impurity radiation of the core plasma. However, the sensor chip is significantly affected by the interference from hard X-rays on, resulting in a significant amount of single-pixel noise in the collected spectral image data. This interference presents significant challenges for accurately analyzing the impurity radiation in fusion plasma. To address this issue, hardware algorithms are prepared to be integrated into the detector to optimize the imaging quality. However, since the existing commercial detector could not be modified in terms of hardware and programming, decision to develop an imaging system for EUV spectral diagnosis. This imaging system will be used on EAST for EUV spectral diagnosis and for future research on hardware optimization algorithms. The imaging system is designed with the GSENSE400BSI sensor chip as the photosensitive component and the AC7Z100B board as the control center, and it also includes two high-speed transmission interfaces: USB 3.0 and Camera Link. Finally, to verify the acquisition function and imaging quality of the imaging system, tests were conducted on the established spectral image acquisition platform. The results showed that the imaging system could normally complete the spectral image acquisition with high- quality imaging.

X-ray Fluorescence Computed Tomography (XFCT) is a molecular imaging technique which is used to reconstruct the distribution of trace elements in samples based on fluorescence signals. However, the quality of reconstructed images is compromised due to sample absorption. In this paper, we propose a deep learning-based XFCT image reconstruction framework to directly transform from the sinogram domain to the image domain, enabling fast reconstruction of XFCT and addressing the fluorescence attenuation issue. Through numerical simulation experiments, it is demonstrated that the Red CNN algorithm improves the NMSE and PSNR evaluation metrics by 0.0249 and 1.3768, respectively, compared to FBP and MLEM methods.

3D hand pose estimation and shape reconstruction have gained considerable attention owing to advancements in deep learning techniques. This task involves estimating hand pose and mesh from images containing hands, with diverse applications across fields such as Augmented Reality (AR) and Virtual Reality (VR). However, accurately estimating and reconstructing complex hand shapes is challenging due to limitations in reconstruction models. To address this challenge, we propose a method that utilizes Graph Convolutional Networks (GCN) to progressively reconstruct 3Dhand meshes, thus improving the algorithm's accuracy and flexibility. This method takes RGB image sequences as input, leveraging the temporal relationships between adjacent frames to estimate 3D pose more accurately from2Dposes. Then, we obtain shallow sparse hand meshes from poses and employ a coarse-to-fine regression strategy to directly regress hand mesh vertices step by step. This approach not only corrects estimation errors in a timely manner, enhancing prediction accuracy, but also reduces the number of vertices regressed simultaneously, lowering hardware computational resource consumption. Our proposed method achieves competitive performance on public hand datasets FreiHAND and HO3D.

Infrared hyperspectral cameras are complicated in structure, require a large amount of imaging data, and the array detector is expensive, which is not conducive to the efficient imaging and real-time data analysis of objects in the infrared regime. Single pixel imaging offers a solution by employing compressed sampling. This paper describes the design and real time operation of a parallel single-pixel near infrared hyperspectral imaging system which can obtain a 128×128×256 spectral cube image with a spectral resolution of 4 nm within the range of 975 to 1736 nm. The compression rate and reconstructed image quality using different orderings of the Hadamard modulation basis, including the natural, cake cutting, and the Russian dolls sequences, are compared. A good reconstructed image of a dynamic object can be realized even with a compression sampling rate as low as 0.78% with the Russian doll sequence. The method proposed in this paper should have many potential applications due to its efficient sampling and high-speed reconstruction enabled by parallel spectral channel computation.

In order to improve the quality of 3D image reconstruction, a 3D image reconstruction method based on correlation imaging is proposed by combining the photometric stereo method with the correlation imaging reconstruction algorithm, and the numerical simulation imaging is carried out. A digital projector (DLP) is used as a light source, and four singlepixel detectors placed in different positions are used to collect signals, and a set of 3D correlation imaging experimental system is built to realize the 3D reconstruction of 256pixel*256pixel target objects in the laboratory environment. This paper provides a new method for 3D image reconstruction and a theoretical basis for the practical application of 3D correlation imaging.

In order to more intuitively judge the relationship between the myopia prevention and control effect of the myopia prevention and control frame glasses and the microstructure parameters of the glasses, this paper designs a saddle surface microstructure array glasses based on the contrast principle, and uses the relationship between the MTF value and the microstructure parameters to establish a quantitative model. The design results show that within the acceptable imaging signal range of the human eye, the saddle surface microstructure array lens can make the light passing through the microstructure unable to converge and image, which greatly reduces the imaging contrast of the retina. When a certain spatial frequency in the range of 0~43lp/mm is selected, the maximum vector height of the microlens is in the range of 0~10μm, and the maximum vector height of the microlens and the MTF value under the maximum off-axis field of view show a nonlinear negative correlation. Therefore, the empirical formula of the maximum vector height and MTF value of the microlens of the spectacle lens is established, and the quantitative analysis of the microstructure parameters and contrast signal of the spectacle lens is completed. This work helps the lens designer to control the contrast control of myopia prevention and control more accurately through the microstructure parameters. At the same time, through analysis, it is found that in the case of relatively small light loss, compared with the spherical microstructure, the saddle surface microstructure has a better effect on reducing the contrast, which is more helpful to reduce the visual quality and slow down the development of myopia.

In remote sensing image processing, image segmentation requires not only the accuracy of segmentation, but also requires the lightweight of the network model as a way to improve the training speed to achieve faster response time. Therefore, in the pursuit of efficient real-time application requirements, this paper improves the lightweight network structure ICNet. Firstly, the efficient channel attention (ECA) module is added to ICNet, and the ECA module improves the feature extraction capability and ensures the simplicity of the model through the weighting operation of the key channel information. Then the joint pyramid upsampling (JPU) module is also introduced, which is integrated into the original ICNet's upper, middle, and lower branching structure for processing feature information. In the subsequent experiments, the proposed EJICNet network structure is trained and evaluated in depth, and the experimental results clearly show that EJICNet significantly reduces the computational complexity while maintaining a higher segmentation accuracy compared to the existing network structure. This proves that the optimisation method proposed in this study balances efficiency and accuracy, and satisfies the network's need for real-time performance.

A single image can not fully describe the information of the target, and the practical application value is low. Fusion of multi-source data to generate images with richer information and higher quality has become a technical research frontier direction in the field of image intelligent processing in recent years. Aiming at the shortcomings of the current photoelectric image and 3D scene image fusion methods, such as poor fusion quality, the existence of artifacts, and the need to manually adjust the parameters for different scenes, in order to obtain more ideal fusion effects and improve the adaptability of the algorithms to the images of different scenes, an adaptive parameter fusion method based on the improvement of HSI is proposed. Firstly, the photoelectric images and 3D scene images are acquired and denoised and aligned; then the statistical characteristics and histogram distribution of the images are considered comprehensively, and adaptive parameterization is used to determine the ClipLimit parameter of CLAHE for image enhancement, and finally the fusion is performed by the improved HSI color model, and the fusion results are tested, and the method improves the quality of the fusion of multi-source aerial images and the fusion effect is significantly better than that of the other fusion methods.

A novel compressive holography imaging scheme is proposed, based on guided filter denoising. In this approach, the gray composite image reconstructed by the traditional algorithm serves as the input for the guided filter. The compressed holographic reconstruction result is then used as a guided image to constrain the input image in the filtering process. Ultimately, a filtered compressive holography reconstruction result with edge-preserving characteristics is obtained through guiding the filtering algorithm. Compared to existing compressed holographic imaging technology, our method effectively retains edge information and significantly suppresses background noise, resulting in high-precision imaging results. The effectiveness of this scheme has been demonstrated through experiments.

Transient sources are important research objects in the field of time-domain astronomy, where nearby celestial bodies or moving objects such as asteroid fragments are suspected transient sources in observations. In the current stage of observation and analysis, the primary method for studying optical transient sources involves capturing multiple consecutive frames of the same sky using a computer and creating a template image using specific algorithms. This template image reflects the state of the sky over a period of time and can be subtracted from any detected image captured at a given moment to produce a residual image. Subsequently, all suspicious target images in the residual image are automatically extracted and presented to the personnel for manual identification, distinguishing between transient source phenomena and false positives. However, this method introduces human factors, leading to low operational efficiency, slow identification speed, and ultimately, the accuracy of identification is affected by the experience of the observer.To address these issues, with the rise of neural networks in recent years, convolutional neural networks (CNNs) can be trained to replace manual identification of transient sources in astronomical images. By feeding continuous astronomical images captured over time into the CNN, the network automatically identifies whether transient sources are captured in the images. Additionally, the CNN can eliminate false positives caused by errors in methods such as image alignment and subtraction. Ultimately, achieving accuracy comparable to human recognition. This project undertakes the following tasks:1. Preparation of training datasets, including two categories: positive samples of actual transient sources and negative samples of false positives caused by errors in data processing. Specific classifiers are trained to distinguish these false positives.2. The main task of this project is to build a classifier based on convolutional neural networks. After establishing the neural network structure based on previous research results, the network structure is further optimized and improved based on the characteristics of the project data, identification requirements, application scenarios, and other factors. The ultimate achievement of the project is to train an excellent transient source classifier, which can achieve an identification accuracy of over 99% when applied in actual observations.

The present single-pixel imaging method usually employs a Hadamard matrix to realize dynamic imaging. When the sampling rate of these methods is reduced, the reconstructed images all appear to have block distortion. This paper proposes a method for single-pixel dynamic imaging using Zernike moments. Five low-order Zernike moments and an appropriate number of high-order Zernike moments are irradiated per frame to obtain positional and image information. The illumination patterns are adjusted according to the acquired data so that they can be used to reconstruct an image of the object at the reference position. It is proved through experiments that the method in this paper can eliminate the block phenomenon and reconstruct precise motion trajectory. The method provides a new optimization direction in the field of single-pixel dynamic imaging.

The analysis of skin lesion dermoscopic images has traditionally relied on supervised learning paradigms, necessitating extensive labeled datasets which can be costly and time-consuming for dermatologists. Conversely, the abundance of unlabeled data offers a promising avenue for self-supervised learning methods in this field. In this work, we introduce a novel strategy for hard negative synthesis that bolsters the efficacy of contrastive learning in skin lesion classification. By strategically down-weighting the contribution of the hardest negative samples during feature-level mixing, our method ensures the neural network prioritizes learning from the most informative and reliable negatives, thereby enhancing the model's feature learning ability. After fine-tuning with a limited set of labeled data, our method demonstrates notable superiority on the ISIC-2016 classification dataset, achieving a 4.29% increase in ROC AUC and a 5.39% increase in F1 score over the standard MoCo-v2 framework.

CCDs (Charge-Coupled Devices) have high sensitivity and low noise characteristics that offer significant ad- vantages for diverse scientific detection applications. However, CCD imaging systems are susceptible to both intrinsic and extrinsic interference due to multiple charge transfers during the readout process. The CCD used in the AST3-II astronomical telescope located at Dome A in Antarctica has been severely affected by electromagnetic interference, resulting apparent stripes across the entire CCD images. To address this issue, we propose a novel correction method based on the Hough transform that effectively mitigates such interference effects. By applying the correction, we can significantly reduce such risks and enhance the reliability of exoplanet detection.

The era of Artificial Intelligence (AI) has brought forth new security threats, leading to the emergence of Reversible Data Hiding (RDH) as a crucial digital technology for safeguarding information security. To tackle the security challenges posed by the transmission of peak points in histogram reversible information hiding, a blind extraction scheme is proposed. Given the minimal differences between neighboring image pixels, peak and sub-peak points tend to be adjacent values, allowing for the embedding of secret information based on these sub-peak points. The sender embeds secret information according to the distribution of sub-peak points within the histogram. As peak points remain unchanged during image transmission, the receiver can accurately identify sub-peak points by referencing them, thus facilitating the extraction of secret information. Consequently, this approach achieves blind extraction. Furthermore, to ensure the security of the information, encryption methods were employed to encrypt the embedded information before embedding. Even if an eavesdropper intercepts the image, they cannot access the private information without the key. The experimental results demonstrate that this method is capable of accurately extracting the embedded information and restoring the image without any loss, thus achieving effective reversible information hiding.

On the basis of the existing road extraction methods and the characteristics of high-resolution remote sensing images, this paper proposes a multispectral-guided hybrid attention mechanism road extraction model (MS-DANET), which introduces a parallel hybrid attention mechanism to process the dual-branch image features between the encoder and decoder structure. It is shown from experimental results that MS-DANET outperforms current road extraction networks and exhibits excellent results with higher road extraction accuracy and better integrity.