Aiming at the problem that the growth state of K2CsSb photocathode cannot be predicted in the current preparation process of K2CsSb photocathode, a prediction model of K2CsSb photocathode reflectivity based on long short-term memory (LSTM) recurrent neural network was proposed. The one-dimensional original reflectivity data set was reconstructed into a two-dimensional data input model after cleaning, screening, serialization and other preprocessing methods. In order to make full use of the highly correlated characteristics of reflectivity data in time series, this model used a double-layer LSTM network to extract features, the prediction results were output through the fully connected layer, and the mean square error (MSE) was used as the evaluation standard for the prediction effect of the model. The experimental results show that the network structure of the model is reasonable and performs well in different data sets, and the prediction accuracy rate can reach 99.21%. The proposed model can be used in the fabrication process of K2CsSb photocathode, and the process parameters can be adjusted by feedback of the reflectivity prediction value to approach the target trend, which can promote the performance of the photocathode.

The high-sensitivity electron bombarded active pixel sensor (EBAPS) not only has the characteristics of high gain and high sensitivity of vacuum imaging devices, but also has the digital characteristics of solid-state imaging devices, which greatly improves the level of night vision imaging. A natural color low-level-light EBAPS imaging system was built based on three-color liquid crystal tunable filter (LCTF). According to the characteristics of low-level-light images, the color enhancement processing such as grayscale stretching, white balance, and color correction was performed on the color images obtained by the system. The experimental results show that the system can realize the natural low-illumination color imaging that reflects the color characteristics of the scene itself, which can effectively improve the observation effect of the characteristics of the target scene at night, and can realize the color imaging under 5×10?4 lx illumination.

In order to realize high-precision monitoring of vehicle emission, a high-precision and wide-range NO measurement method was proposed. Aiming at the problem that the absorption peaks of SO2 and NO in emission overlapping in the UV band, and it was impossible to directly perform single-component gas inversion, the differential optical density (DOD) of mixed gas in the NO sensitive band (200 nm～230 nm) was first calculated by the ultraviolet differential optical absorption spectroscopy (UV-DOAS) method. Then, the adaptive interference cancellation technology was introduced to achieve rapid separation of mixed gas DOD. Finally, the least square method was used to perform the concentration inversion of separated NO. This method could achieve rapid inversion of NO concentration (volume fraction of gas) in the range of 100×10?6 to 3 000×10?6. After testing, the absolute value of the relative error of inversion is less than 10% in the concentration range of 100×10?6 to 200×10?6, and less than 5% in the concentration range of 300×10?6 to 3 000×10?6. This method has the characteristics of large measurement range and fast speed, and can meet the measurement requirements of NO concentration in the range of 3 000×10?6 in vehicle emission.

To solve the problem of stray light in low-level-light optical system detected by low illumination weak signal, the principle and characteristics of stray light were studied. The optical system modeling software LightTools was used to simulate the low-level-light optical system and analyze the stray light. In order to reduce stray light, the extinction threads were processed on the barrel wall of the objective lens barrel, and the energy simulation was carried out for different forms of extinction threads. The simulation results show that the stray light coefficient can be reduced from 7% to 4% by using the extinction thread with the pitch of 0.35 mm, which are consistent with the experimental results, and can provide guidance for the elimination of stray light in the design of optical systems with weak signal detection.

Aiming at the risks that may be encountered in the process of equipment transportation, the photoelectric detectability of four types of consumer-grade surveillance equipment that commonly available in the market (MB, DC, DSLR and DV) was analyzed and achieved effective jamming to each device in combination with transportation safety requirements. By building a cat-eye laser detection system, laser jamming system and echo signal receiving system, it focused on the breakthrough of the theories and key technologies such as detection and identification of typical small-caliber surveillance equipment, and bidirectional imaging between cat-eye detection device and target. On this basis, the detection and imaging jamming effect of the visible-light band laser on cat-eye target under different distances, jamming wavelengths and apertures were analyzed, and an evaluation standard of image jamming effect as well as the effective jamming threshold of different optical surveillance devices were proposed. The experimental results show that the typical optical surveillance device has good photoelectric detectability in the experimental distance and can be jammed effectively. When the laser beam completely covered the aperture of the lens, the closer the distance between the target and the device, the smaller the divergence angle of the laser beam, the closer the waveband to the maximum waveband of the human eye sensitivity (555 nm), and the better the non-harmful imaging jamming effect generated by the laser beam.

In order to solve the problem that single median filtering and gaussian filtering algorithm is not effective in suppressing impulse noise and poisson noise simultaneously in low illumination image, and the edge detail protection is insufficient, an open and close mix-median-gaussian (OCMMG) filtering algorithm based on field programmable gate array (FPGA) was proposed. Firstly, the minimum four-direction difference was used to detect the anomaly degree of each pixel point, the weight was allocated according to the threshold of pulse noise discrimination, and the first step was filtering. Then, the four-direction edge detection algorithm was used to extract image edges, and the second step was filtered according to the set edge confidence characterization value. Finally, the images collected by electron bombarded active pixel sensor (EBAPS) under the condition of 1×10?3 lx illumination were processed by FPGA in real time. The experimental results show that the FPGA processing results are consistent with the software simulation processing results. Compared with the median filtering and gaussian filtering algorithm, the peak signal-to-noise ratio (PSNR) of the algorithm is improved by 3.23% and 16.34%, the structural similarity is improved by 14.66% and 33.86%, and the edge retention index is improved by 0.49% and 4.21%, respectively, which can effectively remove the mixed noise of EBAPS image and meet the real-time requirements.

In order to improve the robustness of crowd counting model to scale and optical noise, a multimodal image fusion network was designed. A statistical model for night crowd was proposed, and a sub network Rgb-T-net was designed. The network integrated the characteristics of thermal imaging and visible image, and the ability of network to judge the characteristics of thermal imaging and night crowd was enhanced. The proposed model used the adaptive Gaussian checking density diagram for regression, and the night vision training and testing were completed on the Rgb-T-CC data set. Through verification, the average absolute error of the network is 18.16, the mean square error is 32.14, and the recall rate of target detection is 97.65%. The counting performance and detection performance are superior to the current most advanced bimodal fusion method. The experimental results show that the proposed multimodal feature fusion network can solve the counting and detection problem in night vision environment, and the ablation experiment further proves the effectiveness of parameters of the fusion model.

Aiming at the problems of low image brightness, poor contrast, and blurred edges caused by insufficient illumination and noise during low-level-light imaging for multi-pixel photon counter (MPPC), an adaptive low-level-light image processing algorithm based on sub-window box filtering was proposed. To reduce the algorithm running time while highlighting the edge detail information of the image, the sub-window box filter was used to layer the image to obtain the basic layer and detail layer. For the image of basic layer, the adaptive threshold histogram equalization was used to stretch the contrast, and the image of detail layer was enhanced by adaptive gain control method. The fusion coefficient was determined adaptively based on the ratio of the number of effective gray values to the total gray in the image of basic layer, and the image of basic layer was fused with the image of detail layer to obtain the enhanced image. Three sets of low-level-light environments with different illumination levels were set by the low-level-light experimental platform for experimental simulation, which verified that the algorithm obtained better results in maintaining edge information and enhancing details. Experimental results show that the proposed algorithm is superior to the previous algorithm in objective evaluation of standard deviation, information entropy, and average gradient, which improves the imaging effect of low-level-light image.

When the electron multiplier (EM) works in a pulse condition, the discrete signals are output on its anode. Since the electron multiplication process has a certain statistical law, to study the performance parameters of the EM in a pulse condition, the substantive tests and analyses of the pulsed signal output from the anode were required. The photomultiplier tube (PMT) based on a dynode EM was taken as an example, by changing the incident light intensity to make it work in a pulse condition, a high-bandwidth and high-sampling-rate oscilloscope was used to collect its anode output signal. A graphical data analysis software was developed based on Python to perform the offline analysis on a large number of pulse signal data collected by an oscilloscope, and performance parameters such as charge integral spectrum, gain, resolution, post-pulse rate, and leading-edge time of PMT could be obtained. The software adopted a modular structure, and each module could work independently according to the different test requirements. The proposed software can quickly analyze the performance parameters of the EM in a pulse condition, and provides a convenient analysis method for the optimization of the EM fabrication process and its application in the field of weak signal detection.

Progressive gain is one of the important parameters for evaluating the performance of dynode photomultiplier tube (PMT), but there is only one related test system at home, which relies on manual operation and the test efficiency is not high. In order to improve the efficiency of testing progressive gain, an automatic test system for the progressive gain of dynode PMT was designed. The continuous adjustment of the light source output intensity was achieved by an automatic light system composed of controllable disphragm, electric baffle and etc, and the on-off of the progressive voltage was controlled by the high-voltage module, the voltage divider module and the relay module. Finally, based on the galvanometer module with an accuracy of 0.01 nA, the acquisition, processing and transmission of the signal was completed, so as to realize the automatic measurement of the progressive gain. The experimental results show that the proposed system can effectively measure the progressive gain characteristics of the dynode PMT, and the test repeatability is within 2%, which meets the test requirements.

The low-level-light (LLL) night-vision viewer is a necessary device in night fighting of information warfare, which is widely used in the field of night-vision individual reconnaissance, gun sight as well as vehicle-mounted and airborne field. Resolution is an important parameter to evaluate the detection ability of LLL night-vision viewer, and also is the key indicator to reflect its comprehensive performance. In order to solve the difficult problems of LLL night-vision viewer in each link of the whole life cycle, such as field operation, maintenance and service, a portable LLL night-vision viewer resolution test system was developed, which was composed by light source components, battery, target wheel, objective lens, illuminometer, voltage measurement feedback module, adapter and touch screen, etc. The proposed test system realized the test of resolution for LLL night-vision viewer, and could replace the laboratory test system on site to make up for the defects such as large size and inconvenient to carry.

The time characteristics of fluorescent screen is one of the important parameters to evaluate the performance of image intensifier. At present, there is no measurement method for the afterglow time of nanosecond fluorescent screen of low-level-light image intensifier. Based on the traditional test scheme of image intensifier afterglow time, a afterglow time test system for nanosecond fluorescent screen was developed. This system used a high-speed signal generator with the sampling rate of 250 MHz to complete the excitation of the laser diode light pulse, and a photomultiplier tube was used with the descending time of 0.57 ns to complete the photoelectric conversion of the fluorescent screen light signal. The weak photocurrent signal of μA magnitude was amplified and converted to a single-terminal differential circuit to complete the AD conversion in AD9684. Then the digital luminance information of the fluorescent screen was stored in the double data rate SDRAM (DDR) unit after field programmable gate array (FPGA), and the host computer sent instructions to read the DDR memory. The USB3.0 high-speed transmission protocol was used to transmit data to the host computer. In the data processing, the Kalman filtering and fast finding falling edge algorithm were used to realize the accurate measurement of noise filtering from collected data and afterglow time. The test results show that the proposed afterglow time test system for nanosecond fluorescent screen can effectively test the image intensifier with ultrafast optical characteristics. The afterglow test results of P47 phosphor reaches 118. 094 4 ns, and the repeatability reaches 2.08%.

The spectral radiance of vacuum ultraviolet (UV) target is of crucial significance to plenty of researches including deep space exploration and spacecraft damage test. Two types of test systems and methods for vacuum UV light sources were studied. Through the research of test method for vacuum UV spectral radiance, a corresponding test system was developed, which included vacuum UV standard light source, optical imaging system, light splitting module, vacuum UV detector module, vacuum chamber and data processing system. According to the direct measurement method and the comparative method, the spectral radiance of deuterium lamps was tested and analyzed, the influencing factors and the relative index error of vacuum UV spectral radiance were discussed, and the accurate measurement of vacuum UV spectral radiance in the five wavelength ranges of 121.2 nm, 135.6 nm, 160 nm, 180 nm, 200 nm at 0.01 μW/cm2·nm·sr~1 μW/ cm2·nm·sr was realized. The repeatability of the measurement is 0.001 34, which shows that the proposed test system can realize the test of vacuum UV signal.

According to the black spot detection principle and black spot halo characteristics of the microchannel plate, a detection method was designed. The circle extraction technology was used to realize the radius and circle center extraction of the fluorescent screen image, and the gaussian Laplacian was used to realize the coarse detection of the target black spot. Based on the characteristics of black spot presenting halo in the periphery at different voltages, the black spot halo was extracted by using the histogram equalization with restricted contrast and combined with the threshold segmentation of the halo detection method. The test results show that the proposed method can effectively distinguish the fluorescent screen scratches, dust and black spots of microchannel plate, so as to realize the automatic extraction of black spots.

The low-level-light image intensifier is an important photoelectric detection and imaging device in the field of night vision, and it is widely used in head-mounted display, hand-held observation and night aiming. The super-second-generation inverted image low-level-light image intensifier with fiber-optic image inverter is widely used in the night aiming system. According to the requirement of practical application, the influence of turning angle error and snake-shaped distortion in the inherent error of inverted image low-level-light image intensifier on the measurement error of aiming and stroke difference in application was analyzed. The analyses show that when the image turning angle error is less than 30' and the snake distortion is less than 30 μm, it can meet the requirements of close-range aiming and measurement of stroke difference. A detection device was developed to control the error, which solved the problem and put forward the relevant quality control points for the production and manufacture of image intensifiers for this kind of application.

As a key electron multiplier, the micro-channel plate is widely used in many fields. The effect of NiCr film as the input electrode of micro-channel plate on the opening area ratio of micro-channel plate was analyzed. A theoretical model was established, and the effects of coating thickness, coating depth and other parameters on the opening area ratio were calculated. Two kinds of coating research were carried out to reduce the loss of opening area ratio. One way, the process was adjusted to weaken the fractionation effect of alloy evaporation, reduce the resistivity of electrode film layer, and the loss of opening area ratio was reduced by about 2 percentage points. Another way was to change the coating method, the Ni and Cr metal were used to prepare laminated films, regulate the proportion of Ni and Cr in the coating process, increase the proportion of Ni in the input terminal electrode, and also reduce the resistivity of the electrode film. On the premise of meeting the surface resistance requirements, the input terminal film could be thinned to 86 nm. Compared with the conventional Ni-Cr alloy film with a thickness of 300 nm, the loss of the opening area ratio of the MCP input terminal was reduced by 3~4 percentage points, and the MCP gain was increased by 6%.

The imaging detectors based on the electron multiplication charge splitting anode of micro-channel plate are often used for weak signal detection such as planetary atmosphere and airglow. In view of the problem of detector imaging distortion caused by unequal gain of detector readout circuit, the theoretical analysis combined with MATLAB simulations were used to reveal the formation mechanism of different distorted images, and a gain compensation method of detector readout circuit was proposed to reduce the detector imaging distortion. The MATLAB simulation and experimental results show that the proposed method can effectively reduce the detector imaging distortion due to the unequal gain of readout circuit.

Ion barrier film is an important part of the third generation low-level-light image intensifier, and its quality plays an crucial role in lifetime and field of view of image intensifier. The effect of ion barrier film quality on the field of view of image intensifier was studied. The ion barrier micro-channel plate (MCP) with typical defects was selected to analyze the field of view of ion barrier MCP at runtime and the quality of ion barrier film, and the effect of ion barrier film defects on field of view of image intensifier at runtime was obtained. The reason for the defects generated in the preparation process of ion barrier film was analyzed, and the solutions to defects were preliminarily put forward, which had a very positive effect on the subsequent preparation of high-quality ion barrier film.

Electron bombarded active pixel sensor (EBAPS) is a hybrid vacuum-solid low-level-light device. Its performance and service life depend on the vacuum maintenance of the device to a certain extent. The reasons for the decline of vacuum degree of EBAPS devices were analyzed, the serious consequences of the deterioration of vacuum degree were deduced, and the means to improve and maintain the internal vacuum degree of EBAPS devices were put forward. By constructing an ultra-high vacuum degassing system, the degassing characteristics of the core components of EBAPS electronic sensitive complementary metal-oxide-semiconductor (CMOS) components were studied. According to the research results, the optimal degassing process parameters were obtained, which provided a technical basis for the preparation of EBAPS digital low-level-light devices.

According to the requirements of high repetition frequency, fast edge speed and small pulse width for cathode gating signal by range-gated technology, a cathode high repetition frequency gating circuit using period and multi-stage acceleration was proposed. By combining the RC circuit and the high-speed gate circuit, the time bias circuit unit was cascaded to generate logic pulses with different time sequences, which could respectively control the intermediate stage drive MOSFET to generate three phased drive signals, and the output of the intermediate stage drive was used as input of the output-stage MOSFET to control the acceleration and retention of its on-off process. It was verified by software simulation and board-level test. The test results show that the proposed gating circuit can increase the edge time of output pulse from μs level to 2 ns, and can provide +50 V/?200 V cathode off/on voltage, so as to achieve a repetition frequency ranging from 0~350 kHZ, a duty ratio of 0~100%, a minimum pulse width of 3.7 ns, and a pulse output delay time jitter of about 0.1 ns. It has important guiding significance for improving the minimum pulse width performance of high-speed and high-voltage gating power, the highest working repetition frequency and reducing the power loss of the device.

Cesium iodide film has the high photoelectric conversion efficiency for ultraviolet light and X-ray, but it is prone to deliquesce in air. The preparation and protective effect of the anti-deliquescence ultra-thin protective film of cesium iodide film on micro-channel plate (MCP) were introduced. The coating depth and thickness of cesium iodide thin film photocathode MCP were tested by scanning electron microscope (SEM). Alumina was used as the protective film of cesium iodide thin film photocathode, and the alumina protective films with thickness of 2 nm, 5 nm and 10 nm was prepared respectively. After storing in the air for different times, there was no obvious deliquescence change on the surface of the cesium iodide thin film photocathode MCP. The gain is about 8 800, and the dark count rate is about 4.1 counts·s?1·cm?2. It is proved that the alumina can be used as the anti-deliquescence ultra-thin protective film of cesium iodide film on MCP.

A high-power YAG laser welding machine was used to seal the back end of the high performance third generation image tube (4J34 kovar alloy) and the fluorescent screen ring (4J49 kovar alloy). The effects of laser power and pulse width on welding joint forming and surface thermal diffusion were studied. The results show that the surface forming quality of 4J34 alloy and 4J49 alloy is the best when the maximum operating current is 100 A, the laser power is 195 W and the pulse width is 1.7 ms. Compared with the laser power, the pulse width has a more significant effect on the weld width and weld depth. The hardening in the welding center is the most serious, and the hardness is the lagest, followed by the heat affected zone.

The phosphors is the key material of the fluorescent screen of low-level-light image intensifier, which can transform the electronic images into the visible optical images. Its performances have an important influence on the resolution, luminescence spectrum, modulation transfer function and afterglow of the image intensifier. In view of the current development of image intensifier at home and abroad, a comparative performance study was carried out on the commonly-used P20 ((Zn, Cd) S: Ag), P22 (ZnS: Cu, Al), P31 (ZnS: Cu), P43 (Gd2O2S: Tb) and P45 (Y2O2S: Tb) for fluorescent screen of image intensifier. The phase structure, spectral characteristics, luminous efficiency and resolution of these five types of phosphors were characterized respectively, and the applicable conditions of different kinds of phosphors were analyzed. The results show that these five types of commonly-used phosphors contribute to the improvement of different performances of image intensifier. Among which, the P43 phosphors has higher resolution, the P22 phosphors has higher luminous efficiency, and the P45 phosphors provides the best observation comfort for the human eyes. In view of the requirements for the high performance of the low-level-light image intensifier, the P22 and the P43 phosphors with superior comprehensive performance can be selected when selecting phosphors for the low-level-light image intensifier, and the appropriate types of phosphors can also be selected according to the specific performance requirements and actual use requirements of image intensifier.