Pixel-level analog-to-digital converters are widely used in infrared, visible and terahertz imaging and other fields due to their unique integrated position with high frame rate, large dynamic range, low power consumption and low noise. In this paper, the principle of mainstream pixel-level analog-to-digital converter structure was introduced; the research progress of pixel-level analog-to-digital converter in dynamic range, power consumption, area and noise and other performance parameters were analyzed; finally several key technologies in each improvement direction and makes prospects were compared and summarized.

A multiple resonance peaks terahertz metamaterial sensor based on double-cross structure was designed. The overall structure was composed of surface metal pattern, intermediate medium and bottom metal plate. The surface pattern of the basic unit was double-cross and the medium material was polyimide. The simulation results show that the proposed structure has five different resonant absorption peaks, and the absorption rate is more than 90%. The first three resonance peaks correspond to the basic magnetic resonance formed in different regions of the unit structure, and the last two resonance peaks are derived from the interaction between the unit structures. When the surface of the sensor is covered with an analyte with a thickness of 15μm, the sensitivity of each resonance peak of the sensor is 72, 137, 234, 353 and 252GHz/RIU, respectively. The resonance peak offset has a strong linear relationship with the refractive index change of the analyte to be measured, which has superior sensing characteristics. In addition, the stability of the multiple peaks sensor is better than that of the traditional single peak or double peak sensor, which has good anti-interference ability to reduce the measurement error. The proposed high sensitivity multiple resonance peaks sensor has a broad application prospect in the field of trace molecule detection.

Optoelectronic oscillator is a new type of microwave frequency source with optoelectronic combination. It can realize signal output with ultra-low phase noise by using optical long-term energy storage. In this paper, the influence of scattering noise in optical fiber on the phase noise of optoelectronic oscillator was studied. The structure of suppressing Brillouin scattering noise based on equivalent broadening of laser line-width by phase modulation was emphatically introduced. Through theoretical formula derivation and experimental verification, it is shown that the above structure can greatly improve the phase noise of optoelectronic oscillator. In the experiment, a phase modulation signal with a modulation frequency of 50MHz and a modulation amplitude of 3.1 is used to broaden laser line-width, and an ultra-low phase noise signal of -157.3dBc/Hz@10kHz at a frequency of 10GHz is obtained.

Theoretical analysis of the self-feedback injection locking of the external cavity of the semiconductor laser was carried out. The influence of the self-feedback injection locking of the on-chip microcavity on the output linewidth of the DFB (distributed feedback) laser was studied, and the key parameters determining the locking bandwidth and the linewidth compression factor were analyzed. Based on the back Rayleigh scattering of the on-chip Si3N4 microcavity with a Q value of 2.4×106, the self-feedback injection locking of the DFB laser was realized. The line width is narrowed from 556.71kHz during free-running to 92.28kHz, and the locking bandwidth reaches 425MHz. The research results contribute to the understanding of the self-feedback injection locking mechanism of semiconductor lasers and provide a new solution with simpler structure and higher integration potential for the realization of narrow linewidth lasers.

The current three-band terahertz filter have a problem of large difference in passband bandwidth. In order to improve the bandwidth uniformity of terahertz filter, a frequency-selective surface based terahertz filter is designed, which consists of three nested "cross" ring metal patches, "cross" metal patches and polyimide substrate. When the terahertz wave is vertically incident on the filter surface, three filter passbands are formed in the range of 0.1~0.862THz, and the peak transmission coefficient of each passband is about 0.9, which has good passband performance, and the transmission coefficient of transmission zero between the passbands is below 0.07, and the out-of-band suppression is good. The maximum bandpass ratio of 3dB bandwidth is about 1.29, and the maximum bandpass ratio of 10dB bandwidth is about 1.13, which has good passband uniformity. The rectangular coefficients of the three working passbands are greater than 1.7, the passband edges are steep, and the frequency selectivity is good. The filter has the characteristics of high transmission coefficient, small bandwidth difference, simple structure and easy processing. It has broad application prospects in the field of terahertz multi-channel communication.

The low half-wave voltage electro-optic modulator is crucial for achieving large-scale photonic integration. This article proposes a thin film lithium niobate Mach-Zehnder electro-optic modulator with a half-wave voltage below 1.5V. The design is based on a single crystal thin film lithium niobate material on an insulator. The influence of the straight waveguide, multimode interference coupler, bent waveguide, and modulation arm on the electro-optic modulator is analyzed. The results show that when the length of the modulation arm is 3mm, this thin film lithium niobate electro-optic modulator has a low half-wave voltage of 1.05V, a low loss of 0.319dB, and a high extinction ratio of 27dB. At the same time, the modulator has a half-wave voltage-length product of 0.315V·cm, high modulation efficiency, and a half-wave voltage compatible with CMOS technology, which is beneficial for large-scale photonic integration.

A new simple method to decouple a 1×2 MIMO dielectric resonator antenna (DRA) by loading a dielectric sheet (DS) on the upper surface of the dielectric resonator (DR) at the lateral side is proposed. The 1×2 MIMO DRA adopted a double-layer dielectric substrates structure to optimize impedance matching and radiation characteristics. The edge-to-edge distance between the two DRs was 0, and the antenna operated in the millimeter wave frequency band. The loaded DS caused the E-field within the DR to redistribute and concentrate towards the DS loading area and within the DS, thereby weakening the field strength coupled to another DR unit to achieve decoupling effect. The simulation results based on ANSYS HFSS software show that the -10dB impedance matching bandwidth of the antenna is 25.6% (22.75~29.43GHz), and a maximum isolation enhancement about 30dB is achieved within the impedance matching bandwidth.

A four-end beam structure piezoresistive accelerometer with high sensitivity and small size is designed in this paper. The structure with different cantilever beam and mass block sizes is modeled using finite element software, and sensitivity and stress analysis, modal analysis, and dynamic response are performed on the structure. The simulation results show that the axial sensitivity reaches 7.40μV/g, the lateral sensitivity is 0.33%, the linearity is 0.06%, the response time is 20.3μs, and the natural frequency is 111.2kHz within the 50000g range. The impact resistance can reach 334000g, which can satisfy the test needs in a high-g environment, and this research provides an accurate and efficient simulation method for the development of high-performance sensors.

The influences of the process conditions on Ag/p-GaN ohmic contact have been studied. An optimized process was obtained and applied in the fabrication of devices. Ag patterns were prepared by three kinds of processes: direct lift-off, lift-off after an oxygen plasma treatment, and wet etching. The adhesive and electrical properties of the samples which were fabricated by the three kinds of processes were compared. It was found that the Ag layer is easy to peel off after direct lift-off process, and the ohmic contact could not be formed after an oxygen plasma treatment. In contrast, the wet-etch samples showed better adhesion and electrical properties after annealing. The mechanism of the influences of different processes on the contact performances were analyzed by X-ray photoelectron spectroscopy. Furtherly, surface chemical treatments before Ag deposition were compared and optimized. The results showed that acidic or alkaline solution treatment could effectively reduce ohmic contact resistivity, and the effect of acidic solution treatment was slightly better. The optimized ohmic contact process can be applied to fabricate the visible light and deep ultraviolet (DUV) LED devices. At 40A/cm2 current, the voltages of the fabricated blue LEDs and DUV LEDs were 2.95 and 6.01V, respectively.

When rough grinding the sapphire substrate, it is necessary to use abrasives with large particle and high hardness to shorten the time. Considering the weak flowability and suspension characteristics of large-particle abrasives, the distribution and motion patterns of abrasive particles during rough grinding were studied by monitoring the swing of the removal amount pointer. Research shows that the dynamic balance of the abrasive layer has a decisive impact on thickness uniformity and surface quality of the substrate, and the establishment of this balance depends on a dense and uniform initial abrasive layer. The supplementary abrasive during rough grinding could only maintain the existing balance and could not be remedied in an imbalanced situation. The work helps to improve the control level of rough grinding process and has certain enlightening significance for in-depth research on the movement of abrasive particles during the grinding process.

A series of red-orange luminescent Y2-2xMgTiO6∶2xSm3+(YMT∶2xSm3+, 0≤x≤0.11) phosphors were synthesized by sol-gel method. The samples were characterized and analyzed by powder X-ray diffractometer (XRD), scanning electron microscopy (SEM), photoluminescence excitation and emission spectroscopy. The results show that the as-prepared YMT∶Sm3+ exhibits a pure-phase structure with no impurity phases. The SEM images show an average diameter of about 2~3μm for the particles and good dispersibility without overt agglomeration phenomenon. Under the excitation of a 407nm near-ultraviolet light, the emission spectrum of YMT∶Sm3+ phosphors has three significant emission peaks in the wavelength range of 500~700nm, which are the stronger orange-red light emission at 603nm (4G5/2→6H7/2) and 650nm (4G5/2→6H9/2), and the weaker green light emission at 569nm (4G5/2→6H5/2). The optimal doping concentration of Sm3+ ions is about 5mol%, and the theoretically calculated critical transfer distance of Sm3+ ions is 1.619nm. When the concentration of Sm3+ exceeds 5mol%, there is an obvious concentration quenching effect, which is attributed to the electric quadrupole-electric quadrupole (q-q) interaction between Sm3+ ions. The CIE chromaticity coordinates of YMT∶Sm3+ phosphors are located in the orange-red region, indicating that it has potential application value in red phosphors for white LEDs.

To solve the problems of low efficiency and high labor intensity in manual interpretation of FISH (Fluorescence In Situ Hybridization) fluorescence images, an improved YOLOv5 algorithm that integrates spatial image enhancement is proposed for intelligent cell detection in FISH fluorescence images. On the basis of the original YOLOv5 neural network model, the algorithm added a spatial image enhancement module, and the optimal enhancement coefficient of this module was selected. This module expanded the contrast adaptation range of the model to fluorescence images, and improved the feature extraction ability and cell detection accuracy of the model. The experimental results show that the mAP (Mean Average Precision) of the improved YOLOv5 model is 0.983, which achieves better training performance and convergence speed than the original model. Furthermore, the improved YOLOv5 model achieves a cell recognition rate of 91.65%, which is 9.19% higher than that of the original YOLOv5 model. Embedding the intelligent cell detection algorithm into the self-developed fluorescence image intelligent detection software, combined with fluorescence point detection algorithm, it can give effective interpretation results.

In order to solve the problems of the short girth and the insufficient error-correction performance for quasi-cyclic low-density parity-check (QC-LDPC) codes, a novel construction method for QC-LDPC codes based on the protograph is proposed. In this method, the protograph whose code-length and code-rate could be flexibly adjusted was selected as the base matrix. Then the Lucas sequence and arithmetic progression with special properties were combined to construct the parity check matrix with ring-length of at least 8 through the low decoding threshold of the protograph and the special properties of the sequence, which required less storage space and was easy to be implemented in hardware. The simulation results show that the PLA-QC-LDPC (2400,1200) code constructed by this method shows a certain degree of net coding gain compared to the fast codable irregular LG-QC-LDPC code based on the Lucas sequence and the greatest common divisor (GCD) sequence, the PM-QC-LDPC code based on the prime number and the multiplication table, and the irregular PL-QC-LDPC code based on the protograph and the eliminating basic trapping sets with the same code-length and code-rate.

Satellite laser communication link is a key technology for achieving large-scale constellation networking of satellites. Coherent laser link has high sensitivity, resistance to background interference, and large space for rate upgrade, making it widely used in inter-satellite laser link. In this paper, a signal-to-noise-ratio (SNR) analysis model was established for satellite coherent laser communication terminal with front EDFA. The effects of EDFA power gain, EDFA noise figure, local optical power, signal optical power, optical bandwidth, and electrical bandwidth on the terminal output SNR and the power ratio of various output noise were analyzed by simulation. The SNR characteristics of satellite coherent laser communication terminal with front EDFA were obtained.

In order to improve the positioning accuracy of the current indoor visible light positioning system, an improved particle swarm optimization algorithm considering the dynamic inertia weight and cognitive factors of noise interference is proposed. Firstly, the Euclidean distance that determined the positioning accuracy was transformed into the optimization problem of the minimum value of the objective function. Secondly, the dynamic assignment of inertia weight was used to enhance the global search ability in the initial stage and the local search ability in the later stage of PSO. Then, the value of individual cognitive factors was reduced nonlinearly by sine function, and the value of group cognitive factors was increased linearly by cosine function, which further improved the positioning accuracy. Finally, the proposed localization algorithm was verified by simulation and experimental test. The results show that in the simulation test, in the 5m×5m×5m and 5m×4m×3m positioning models, the average spatial positioning errors of the four height planes of 0, 0.5, 1 and 1.5m are 0.65 and 0.54cm respectively. In the experimental test, the average positioning errors in the 1m×1m×0.8m and 1m×0.8m×0.8m indoor space are 2.67 and 1.81cm respectively.

In high-speed interface circuits, receivers typically use Continuous-Time Linear Equalizer (CTLE) to eliminate the effects of Inter-Symbol Interference (ISI) on signal transmission. In order to improve the high-frequency gain of the CTLE circuit and reduce the chip area, a CTLE circuit with a maximum rate of 50Gbps was designed based on the UMC (United Microelectronics Corporation) 28nm process. Its main circuit was composed of a two-stage CTLE circuit with a trans-admittance transimpedance (TAS-TIS) structure and feed-forward path. On the basis of traditional CTLE, the active inductor was used as the load, the transimpedance amplifier was built based on the inverter, and the feed-forward path was added to the input tube, which effectively expanded the operating frequency of the circuit. The simulation results show that the eye widths of the 40Gbps PAM4 (4-Level Pulse Amplitude Modulation) signal, 50Gbps PAM4 signal and 28Gbps NRZ (Non Return Zero Code) signal reach 0.68, 0.5, 0.92 code element spacing (UI) respectively, which can meet the requirements of the post-stage circuit for input signals. It is of great significance to improve the overall transmission data rate.

In the actual laser detection process of ship wake, there are impurities (bubble groups, suspended particles) between the laser detection system and the target bubble layer, resulting in the reduction of the signal-to-noise ratio of bubble echo signal. In order to improve the signal-to-noise ratio of the actual ship wake detection, Monte Carlo method was adopted to simulate the occlusion of different bubble layer characteristics. By changing the thickness and number density of the occlusion bubble layer, the influence of the occlusion bubble layer on the target bubble echo characteristics was explored. The simulation results show that when the occlusive bubble layer exists, the echo signal decreases obviously, and the decreasing trend is more severe with the increase of the occlusive bubble layer thickness and number density. A simulation ship wake laser detection system was built under laboratory conditions, and the occlusion of different bubble layers was verified. It is concluded that the occlusion effect becomes stronger with the increase of bubble layer thickness and number density. The normalization of test data is carried out to realize the mutual verification of simulation and experiment, which can provide support for the engineering of ship wake laser detection.

Satellite remote sensing technology was used to obtain high-resolution remote sensing image water body data set. Based on deep convolutional neural network, water body was extracted from high-resolution remote sensing images and intelligent monitoring of impaired water body was carried out. An Improved Water Detection Network (IWDNet) model based on the proposed U-Net was proposed. Firstly, based on the U-Net structure, skip multi-scale feature fusion was introduced, and different Multi-scale Feature Fusion Attention Mechanisms (MFFAM) modules were generated by combining SE, ECA and CBAM attention mechanisms for comparison. The dilated convolution was introduced to expand the network receptive field. Finally the recognition and detection of impaired water bodies were realized. Experiment results show that the MFFCBAM-IWNet model based on skip multi-scale fusion and CBAM attention mechanism effectively improves the recognition accuracy, and performs best on the high-resolution remote sensing image water body data set. The overall accuracy is 98.56%, and the Kappa coefficient is 0.9784.

Aiming at the problem of memory consumption in ultra-high resolution image processing and the problem of over-stylization in the process of style transfer, a method of ultra-high resolution image style transfer combined with reversible network is proposed. The algorithm used the reversible Glow module as the basic unit to construct a reversible neural network module, and divided the image into small blocks for processing. In the style transfer module, a residual module with a channel attention mechanism and a thumbnail instantiation normalization module (TIN) were used to ensure that the styles of each module were consistent. A global-local loss calculation method was proposed, which could effectively deal with local structural features. Experimental results show that, compared with the current general-purpose neural style transfer network, this algorithm can not only avoid the information loss problem in the process of image encoding and decoding, but also achieve better style transfer performance at a lower memory cost.

Endmember extraction is the key step of the mixed pixel decomposition in hyperspectral remote sensing images. Traditional endmember extraction algorithms ignore the spatial correlation and nonlinear structure of hyperspectral images, which restricts the accuracy of endmember extraction. Aiming at the spatial relationship and nonlinear structure of hyperspectral images, a nonlinear endmember extraction algorithm based on homogeneous region segmentation is proposed. The hyperspectral image was divided into several homogeneous regions using superpixel segmentation method, and the manifold learning method was used to ensure nonlinear structure of hyperspectral images, extracting preferred endmembers within homogeneous regions. Simulation data and real hyperspectral image experiments show that the algorithm in this paper can guarantee the nonlinear structure of hyperspectral data, and the endmember extraction results are better than other traditional linear endmember extraction methods. In the case of low signal-to-noise ratio, it can maintain a good endmember extraction results.

In order to address the problems of indistinct features, time-consuming field interpretation, and low accuracy in Ground Penetrating Radar (GPR) echo image recognition, an improved method called GSR-YOLOv7 based on YOLOv7 is proposed in this study. First, GhostConv was used to replace the convolutional kernels in the YOLOv7 convolutional layers, thus reducing the number of parameters. Second, the SimAM attention mechanism was introduced to improve the feature learning capabilities. Third, the reuse of the receptive field block (RFB) convolutional kernel was implemented to enlarge the receptive field. Finally, an improved EIoU loss function was employed to improve model classification and regression accuracy. Experimental results on a dataset of road echo images demonstrate the effectiveness of the proposed approach. The GSR-YOLOv7 model achieves a MAP50 score of 97.57% and a MAP50∶95 score of 73.13%, showing improvements of 2.13% and 8.46%, respectively, over YOLOv7. In addition, the model size is reduced by 35%. The GSR-YOLOv7 model exhibits excellent detection performance for echo targets and is suitable for use on mobile systems. It has significant value for platforms with limited processing power and small form factors.

Aiming at the problem of poor detection of vehicles and pedestrians by a single sensor in complex road conditions and bad weather conditions, a set of visible light, visible polarization, short-wave infrared and long-wave infrared multimodal data acquisition system was built to construct a multimodal dataset, and a multimodal pedestrian detection algorithm for vehicle pedestrians was proposesd. Firstly, a heterologous image registration algorithm based on the multi-scale partial intensity invariant features of improved SIFT feature points was proposed. Then, for target detection, a multimodal data object detection network based on YOLOv5 was proposed. Finally, the average accuracy is improved by 1.0% in the daytime dataset and 10.9% in the daytime and nighttime mixed dataset.

In order to address the problems of poor detail resolution and blurred target edges in infrared images, an image enhancement method based on improved Generative Adversarial Network is proposed. Firstly, the generator was constructed based on the codec network U-Net, optimizing the U-Net skip connection method and fusing the global context module to achieve contextual modelling of global and local features. Secondly, the discriminator is constructed based on the Capsule Networks, the capsule network structure is improved by combining with Res2Net structure and the fully connected layer of the Capsule Networks was deconvolutionally reconfigured to achieve multi-scale image feature extraction and reduce model parameters redundancy. The experimental results show that, compared with the current mainstream algorithms, the algorithm in this paper can effectively highlight the detail information, suppress the noise, and improve the image resolution and visual effect.

Due to the complex background information and too much interference information of the optical melanoma image collected by dermoscopy, the detection accuracy is low, and there are problems such as false and missed detection. Therefore, an optical melanoma image detection algorithm based on reparameterized large kernel convolution is proposed. Firstly, in the backbone part, a new module C3_RepLK that combines large kernel convolution and C3 was designed to increase the receptive field of the model and extract more effective information. Secondly, the receptive field module RFB was introduced to fuse the feature information of different scales to reduce the problem of false detection. The hybrid dense sparse convolution GSConv and lightweight upsampling operator CARAFE were used in the neck network, so that the network could capture rich context information and suppress the missed detection problem. Finally, the second-order channel attention module SOCA was integrated into the algorithm to strengthen the correlation between features and focus on more useful features. Experimental results show that compared with the original YOLOv5 algorithm, the proposed detection algorithm improves the average accuracy of all categories from 85.0% to 89.4%, which proves the effectiveness of the proposed algorithm for detecting melanoma.

The laying angle deviation of FBG shape sensors leads to measurement errors in curvature and torsion, which in turn affects the reconfiguration accuracy of shape sensors. To address this issue, this paper proposes a self-calibration model for FBG laying angle deviation based on whale optimization algorithm (WOA). The reconfiguration experiment verified the effectiveness of the proposed method. The remote reconfiguration errors of different shapes were reduced from 11.66 and 22.6mm to 5.63 and 10.47mm, respectively. The relative errors were reduced from 2.56% and 4.96% to 1.21% and 2.25%, and the relative errors were reduced from 2.56% and 4.96% to 0.95% and 2.06%, respectively.

A system architecture that combined 5G power virtual private network and digital twins was designed in this paper, and the construction method of synchronization during the digital twins of terminals was considered. Then, a hierarchical multi-agent reinforcement learning algorithm was proposed to determine the allocation of resources to power user terminals and the placement of digital twins. The upper layer achieved the allocation of sliced communication resources through deep Q network, and the lower level achieved the placement of digital twins in power user terminals through multi-agent deep reinforcement learning. The experimental results show that the proposed hierarchical multi-agent algorithm can achieve better system benefits when the synchronization strength of the digital twin of the power user terminal is reached.

Instantaneous frequency measurement (IFM) is one of great important technologies in modern electronic warfare. The photon-assisted IFM technology was paid particular attentions due to its huge bandwidth, low loss, small size, light weight, as well as electromagnetic inference immunity, which can overcome the bottleneck of traditional electronic methods. Based on the existing research, another photon-assisted instantaneous frequency measurement method with smaller frequency measurement error was put forward, which realized the frequency measurement of 0.5~18.5GHz signals by constructing an optical link with complementary characteristics of radio frequency power response. The results show that the signal gain is consistent with that of theory. Then the amplitude comparison function with more steep slope, is derived to measure the unknown signal frequency, which can improve its accuracy. The experimental results indicate that the frequency measurement error is less than 30MHz. Moreover, this method is very simple and just a few ordinary optical devices are needed. Meanwhile, it has favorable ability to adapt to the environment. By replacing the modulator and photodetector with that of higher operation frequency, it can be easily extended to work for ultra-wide band signal measurement.