In remote sensing imaging, extinction ratio and detector noise are important parameters that determine the accuracy of polarization imaging of nanowire gratings. To strike a balance between the two parameters for system optimization, a mathematical model using the two parameters and polarization noise was established in this paper. The photoelectron number received by large pixels was used as a measure to quantify the relationship between detector noise and system polarization noise. The effect of the two parameters on polarization noise was compared via a simulation wherein the polarization state of incident light was changed. Then, a platform to test the accuracy of polarization imaging with adjustable extinction ratio and exposure time was built to verify the mathematical model and simulation results. The simulation results demonstrate that when the extinction ratio of the system is greater than 20, wherein increasing detector noise by increasing the number of photoelectrons received by large pixels makes a greater contribution to the polarization accuracy than increasing the extinction ratio.

To obtain a faster rate of polarization imaging detection, this study has improved the existing mechanical rotary polarization imaging device and designed a continuously rotating polarizer imaging device. Image processing flow and polarization imaging speed were also improved. This device overcomes the shortcomings of traditional mechanical rotating polarization imaging with a large volume and slow imaging speed. The system uses a motor to control the polarizer to rotate quickly and smoothly, the camera to work synchronously for the quick acquisition of the polarized image. Simultaneously, the polarization image is pipelined to achieve a higher polarization image acquisition rate, and each three adjacent intensity diagrams are used to calculate the polarization image which resulted in the same imaging rate of polarization image and intensity image. Experiments show that the device has good working stability and can complete the acquisition of polarized images well. Moreover, it can obtain polarization information through the polarization and polarization angle with an average acquisition time of 0.033 s. This work improves the detection speed of the mechanical rotary polarization imager and enables the enhancement of the mechanical polarization imaging speed and achievement of the basis of detecting moving targets.

Swept source is a key component of optical coherence tomography; its spectral bandwidth and instantaneous linewidth directly affect the axial resolution and imaging depth of the imaging system. In a single filter, the two parameters are mutually exclusive. Here, a swept source system with two filters combined for optimization was proposed. Two semiconductor optical amplifiers in parallel were used as the gain medium. An acousto-optic tunable filter (AOTF) and a Fiber Fabry-Perot Tunable Filter (FFP-TF) were connected in series in a ring cavity. The tuning range and instantaneous linewidth of the AOTF were relatively wide, while those of the FFP-TF were comparatively narrower. Post synchronization and matching setting, the two filters worked in coordination with mutual advantage. The system delivers a swept output with a center wavelength of 1 316 nm. The spectral range is between 1 235-1 380 nm, while the tuning range, instantaneous linewidth, sweep speed, and output optical power are 145 nm, less than 0.02 nm, 1.35 kHz, and 0.48 mW, respectively. The swept light source can overcome the inherent defects of a single filter and achieve an effective compromise between the broad spectral bandwidth and narrow instantaneous linewidth. This is of great significance for optimizing the overall imaging performance.

A novel ultraviolet-visible-near infrared imaging spectral system in air has been demonstrated to provide an effective instrument to measure the environment and pollution of the ocean along the coastal area. Based on the characteristics of the targets, different parameters of the performance of the system have been proposed. The Dyson imaging spectrometer was chosen to satisfy the requirements of the signal-to-noise ratio and the high optical performance. The Dyson spectrometer has not been widely applied in the engineering field because of its structural limitations. To solve this problem, the distances amongst the slit, the imaging plane, and the optical elements were increased to modify the traditional shape of the spectrometer. Therefore, perfect aberration eliminated conditions of the advanced system were obtained through analysis and addition of lenses. The advanced Dyson spectrometer presents excellent design features with an NA of 0.278 in the range of 320-1 000 nm. The Modulation Transfer Functions (MTFs) in all fields of views and all wavebands are larger than 0.5 at the Nyquist frequency 38.5 lp/mm of the detector. The prototype has a spectral resolution of 3.375 nm, satisfying the design requirement. This new spectrometer is more convenient and effective for coastal ocean observation.

A one-step ultrasonic welding approach for metal-packaged fiber Bragg grating (FBG) sensors was proposed to solve the problem of aging and creep in the field of traditional adhesive packaging. FBGs with and without polyimide coatings were bonded to the surface of an aluminum alloy substrate via one-step ultrasonic welding. The spectrum and thermal and mechanical properties of the bonded FBGs were studied. Additionally, the cross sections of the FBG sensors were analyzed by a Scanning Electron Microscope (SEM). The results reveal that the reflection spectrum of the coated FBG sensor has no obvious distortion, and the side-mode suppression ratio is higher than 10 dB. It is shown that the temperature sensitivity coefficient of the polyimide-coated FBG is 34.63 pm/°C, the strain sensitivity coefficient is 1.18 pm/με, the strain transfer efficiency is 98.5%, and the linearity reaches 0.999, the value of which is higher than that of the uncoated FBG sensor. After performing several repetitive temperature impact tests at temperatures ranging from 14.2 °C to 80 °C, it is found that the uncoated metal-packaged FBG sensor is destroyed, whereas the coated sensor maintains excellent temperature response characteristics. The SEM results show that the metal alloy bonded well to the surface of the coated and uncoated optical fibers. Furthermore, the one-step ultrasonic welding technique was confirmed to not require metallic pretreatment on the FBG surface; thus, the technique is relatively simple and requires less time to implement. The results indicate that the polyimide coating can effectively improve the reliability of metal-packaged FBG sensors for measuring temperature and strain, which would be useful in rigid-environments and for long-term sensing.

Carbon contamination in a synchrotron radiation (SR) beamline is a significant factor affecting the transmission efficiency in the beamline, particularly in a soft X-ray beamline. Currently, the main solution to carbon contamination is to clean or replace the contaminated optical components. A large number of experimental studies have shown that the formation of carbon contamination is related to the presence of a small amount of oil on the chamber wall. Therefore, the purpose of this study was to clean the vacuum chamber wall of the mirror box and obtain a cleaner vacuum chamber by glow discharge, striving to fundamentally prevent the formation of carbon contamination. This paper applied the Glow Discharge Cleaning (GDC) system to the Shanghai Synchrotron Radiation Facility (SSRF) beamline mirror chamber. The volt-ampere characteristics of the device under different vacuum pressures were studied. Mass spectrometry was then performed to analyze the residual gas in the mirror chamber before and after the glow discharge as well as the GDC process. The result shows that the primary cracking products of residual oil molecules in the vacuum chamber were mainly particles with molecular weight of 69. After being cleaned by glow discharge, the quantity of oil molecules (molecular weight 39, 41, 43, 55, 57, 69, 71) can be reduced by 50%, proving that glow discharge has a remarkable effect on removing oil molecules from the surface of the mirror chamber in the beamline. The GDC system is of great significance to reduce carbon contamination, especially in soft X-ray beamlines.

Efficient broadband reflective Extreme Ultraviolet (EUV) multilayers require superior control and precision of layer thickness. A solely time-controlled deposition system can not meet the requirements of high accuracy. In this paper, we present a scheme for the design of broadband multilayers with discrete thicknesses based on an evolutionary algorithm. This method greatly improves the reflectivity curve compared to that of conventional multilayer mirrors without discrete thicknesses. To verify the superiority of the design, the broadband multilayers were deposited using a magnetron sputtering system. The EUV measurements reveal that the deposited aperiodic broad angular multilayers exhibits reflectivity values greater than 41% over an angle of incidence range of 0-17° for a fixed wavelength of 13.5 nm, the broad angular multilayers in four different stacks exhibits reflectivity values greater than 35% with a wide angular bandpass over an angle of incidence range of 0-18.5° for a fixed wavelength of 13.5 nm, and the broadband multilayer mirrors exhibits reflectivity values greater than 21% for wavelengths ranging between 12.9-14.9 nm for a fixed angle of incidence of 3°. This study demonstrates a great potential for the application of discrete design in the fabrication of EUV broadband multilayers with high accuracy.

Yb: LuScO3 crystal is a new type of gain medium used in solid-state lasers. The surface profile and surface quality of the Yb: LuScO3 crystal affect the characteristics of the output laser beam significantly. Therefore, it is extremely important to explore the processing parameters for its ultraprecision optical manufacture. In this paper, a systematic study of the processing parameters for the optical manufacture of Yb: LuScO3 crystal was reported. To solve issues related to the brittleness of the Yb: LuScO3 crystal and the poor quality of the generated surface, the key technology of stitching and the use of copper resin pads were proposed. First, the stress due to different protective paddings was simulated using COMSOL Multiphysics software and the size of abrasive B4C particles was continually decreased during the stitching and lapping stages, respectively. Next, copper resin pads were used during the stage of rough polishing, and their function was explained. Finally, the output power of the continuous wave laser was achieved by diode-pumping the finely polished Yb: LuScO3 crystal. The results reveal a final surface roughness of 0.296 nm (root mean square value) and surface accuracy of 53 nm (peak-to-valley value). An output power of 8.3 W and a slope efficiency of 58% were obtained using a diode laser pump source at a wavelength of 1 086 nm. This method can be widely used for the high-precision machining of Yb: LuScO3 crystals.

Multilateration with laser tracker is an important method for on-site traceability in large-scale equipment manufacturing processes. An accurate evaluation of uncertainty is the key to ensuring quantitative accuracy in manufacturing processes. In this paper, we reported an accurate and fast technique for the evaluation of uncertainty in multilateration with laser tracker. First, the sources of uncertainty in multilateration were analyzed; these primarily include instrument errors, environmental conditions, and manufacturing errors of the target. Next, based on the propagation of uncertainty for multivariate measurement error models (GUM), the uncertainty in multilateration with laser tracker was estimated. Finally, the uncertainty in the point-to-point length was calculated. Our experiments reveal that the differences between the uncertainty values evaluated using GUM and Monte Carlo method (MCM) are within the numerical tolerance limit. The deviation in the uncertainty in coordinates is less than 0.000 2 mm, whereas that in the correlation value is less than 0.01. The time required for the GUM method is only 0.08% of the time required for the MCM method, and the value of En obtained from the point-to-point length test is less than 1. Thus, this study reveals that the evaluation of uncertainty in multilateration with laser tracker based on the GUM method is feasible and efficient, and the obtained results are accurate and reliable.

The precision cable drive is a flexible frictional transmission method in which power is transmitted from drive capstans to a driven pulley by a properly preloaded transmission medium. Owing to its many advantages, such as lightweight, high efficiency, and simplicity, it has been widely used in the field of lightweight electro-optical tracking. Cable tension is an important design parameter that can effectively regulate the transmission characteristics of the precision cable drive. However, accurate measurements of cable tension in highly confined spaces are difficult to make with the existing tension meter method. Therefore, a revised three-point-bending method was proposed in this paper. Parametric sensitivity was investigated to evaluate the relationship between the transmission characteristics of the drive and cable tension. Furthermore, a revised model for measuring cable tension that included bending rigidity was established. To validate the proposed theoretical models, a series of experiments were performed with two types of cables and three kinds of measuring spans. The test results reveal that the tension deviation could can be reduced to 5 N by using the proposed method, which is considerably lower than that obtained with the tension meter method reported in the literature. The revised method can meet the requirements of cable tension measurement in a precision cable drive system.

The methods for obtaining the solutions to kinematics error with nonparametric constraint and dynamic estimation were proposed herein to estimate the position and orientation of a manipulator in the entire workspace domain. The pose error model of dynamic nonparameterized constraint of manipulator was constructed based on the equivalent differential transformation of the link system, which in turn was based on the error equivalent differential variable and multi-joint motion. This study presented a dynamic functional description of the comprehensive error of the manipulator as the periodic dynamic function related to the joint angle variable was changed. Online decoupling transformation and compensation algorithm of a multi-joint motion space coordinate system were designed in accordance with the coupling law of joint motion of multilink coordinate systems. Verification experiments performed in the entire workspace domain reveal that the error estimation and measured values of the coordinates between the positions have maximum absolute deviations less than 0.01 mm, while the absolute deviation of the orientation angle is less than 0.03°. The experimental results indicate that this method can improve the accuracy and reliability of error estimation in the entire workspace domain.

In order to realize the uncertainty evaluation of the workpiece circularity error, the sampling strategy, error evaluation method, and uncertainty of the circular outline of the workpiece were investigated based on the Coordinate Measuring Machine (CMM). First, to achieve many samples from different workpieces, circular outlines were measured. Next, the sample circularity errors were evaluated according to the Differential Evolution (DE) algorithm. Then, by comparing the errors, the adopted sampling strategies and the DE algorithm were explained. Finally, based on the results of the circularity error, the uncertainty was evaluated by applying the GUM and MCM methods. The maximum average difference is 1.1 μm, and the average standard uncertainty of the MCM method is 0.02 μm less than the GUM method. More stable, reliable, and accurate results can be obtained using reasonable sampling points, DE algorithm, and MCM evaluation method.

To analyze the effect of a variable section throttle on the performance of aerostatic bearings, an aerostatic bearing model with a variable section throttle was proposed. Furthermore, the dynamic change in the shape of the throttle cavity cross section was realized by deformation of the elastic plate of the bearing surface. First, the coupled partial differential equations of solid plate deformation and gas lubrication were established. Then, they were discretized and solved by the high- precision finite difference and over relaxation iteration methods. The results reveal that the shape of the throttle determines the value of the nozzle coefficient in the numerical calculation. Furthermore, the stiffness of aerostatic bearings with a variable section throttle is 15% higher than that of aerostatic bearings with a rigid throttle, indicating that a variable section throttle allows aerostatic bearings to achieve a greater stiffness under high bearing capacity. The results of the theoretical analysis are in good agreement with experimental results. Moreover, they indicate that a variable section throttle can effectively improve the static characteristics of aerostatic bearings.

To realize high precision position control of a gene sequencer motion stage, a gene sequencer motion stage control system was developed, and its applied methods such as mathematical modeling, model identification, controller design, and input shaping were investigated. First, based on the dynamic equation of the motion stage and the voltage-force relationship of a permanent magnet linear synchronous motor, a model of the motion stage was established. Then the method of frequency domain scan was adopted to determine the model parameters. Finally, a compound controller, combined with a double closed loop controller and forward feedback controller, was designed based on the model of motion stage to ensure stability and high precision of the motion stage, and an input shaper was designed based on the dominant pole of the compound control system to eliminate oscillation in the motion stage. Experimental results indicate that the compound controller with the input shaping improves the repeated positioning accuracy of the motion stage from ±1.47 μm to ±0.354 μm. The proposed design allows the motion stage to achieve the ultimate repeated positioning accuracy faster than the conventional compound controller, and satisfies the requirements of high stability and precision when the gene sequencer generates an image.

To investigate the temperature rise caused by loss of the magnetically suspended control moment gyroscope for the spacecraft application, it is necessary to analyze and calculate the loss and temperature distribution. In this paper, theoretical loss models that consider the iron and copper losses were established. The losses in a single gimbal magnetically suspended control moment gyroscope (SGMSCMG), which consists of a frame torque motor, radial magnetic bearing, axial magnetic bearing, high-speed motor of rotor system, with a rated speed of 12 000 r/m and maximum angular momentum of 200 N·m·s were calculated. These losses were then used to determine the temperature distribution, which was based on a three-dimensional finite-element model. A thermal-structure coupled simulation analysis was also performed. The results reveal that the maximum temperature occurs at the stator of the high-speed motor and has a value of 48.3 ℃. Finally, the temperature rise of the prototype is verified by experiment. According to the experiments, the maximum temperature occurs at the stator of the high-speed motor, and its maximum value is 51.8 ℃. The error is 6.8% when the experimental results are compared with predicted values. The loss calculation and finite element analysis of the thermal field are verified by the temperature rise experiment. The experimental results provide theoretical reference for overall structure optimization.

This study investigates the influence of thermal history (molding temperature and cooling rate) on the mirror forming process for spin-cast aspheric mirrors. Abaqus finite element analysis software using different molding temperature and cooling rate of simulation analysis to determine the Spin-cast temperature is 950 ℃ and the cooling rate of mirror is 1 ℃/s which can be completed in the mold filling and mirror forming. The spin-cast experiment was executed, and the forming results were measured. When the cooling rate is more than 1 ℃/s, the mirror is fragmented owing to the excessive local residual stress after cooling. Additionally, the axial height of the mirror edge differs from the theoretical value by 10.12 mm, and there is a large surface figure error. This surface figure error is attributable to the volume change on cooling.

Piezoceramic actuators are widely used in precision positioning and control; however, their asymmetric hysteretic characteristics severely affect the position and control accuracy of a system. To address this problem, a modeling method was proposed based on the generalized Bouc-Wen model, and the system parameters of the model were identified using the differential evolution method. Based on the generalized Bouc-Wen model, a hysteretic compensation control strategy with an analytical form was developed and an internal model control scheme to control the piezoceramic actuators was proposed. An experimental platform was developed to verify the effectiveness of the modeling and control strategy. The results of modeling a piezoceramic actuator reveal that all modeling errors are within 0.051 0. Compared with the conventional Bouc-Wen model, our proposed control model can reduce the modeling errors by approximately 21%-46%. Experimental results from the tracking of amplitudes of 20 μm and frequency signals within 100 Hz indicate that the proposed control method offers effective real-time tracking performance and control accuracy. For 100 Hz, the root mean square error and relative error between the reference and output of the piezoceramic actuators were 0.491 6 μm and 0.040 2 μm, respectively, indicating that the proposed control model can satisfy the requirements of practical applications.

Considering large optomechanical structures, a fluid damper was designed with high damping and low axial stiffness in the full frequency range to reduce the width of frequency noise. Firstly, the theory of parameter design for a liquid damper was reviewed. Secondly, the characteristics of the fluid damper were verified using the finite element method, in addition to its influence on the entire structure. The simulation results indicated that the use of a liquid damper could effectively increase the structural damping of a spacecraft without affecting its mechanical characteristics. Moreover, it was shown that the line of sight of an optomechanical structure can be improved by more than 50% generally. The test system used to study the characteristics of the design parameters of the dampers was designed as part of this investigation. It was determined that the damping of the liquid damper decreased with an increase in frequency. A coefficient of more than 300 N·s/m was obtained at 300 Hz, while the change of the stiffness with frequency remained approximately the same. These results indicate that the experimental data are consistent with the simulation results, and the design of the stiffness and damping properties of liquid dampers satisfied the requirements. In summary, the effectiveness of liquid dampers on vibration suppression of large optomechanical structures was verified based on simulation and test results.

For the assembly requirements of special satellite components, robot assembly technology based on visual guidance and force feedback control was studied. This technology gives the robot flexibility under different working conditions and offers high application efficiency under variable satellite assembly conditions. A robot assembly scheme that combines visual and force information was presented. Auxiliary pins were installed in mounting holes, a component was guided to the taper area of the pin by the robot using a visual guide, and a force feedback control was applied to the robot. The component was then accurately positioned based on the pin guidance. Infrared cameras and cooperative targets were used to achieve stable visual recognition and target positioning. A probe-type measurement tool was designed and measurement methods were developed to achieve flexible and convenient measurements of target points. A method for calculating the pose transformation matrix and target position of a robot was proposed based on known spatial correspondence point pairs and was used to achieve compliant pin guidance control. Experimental results show that the measurement matching error of the corresponding hole was within 2.9 mm. The robot can convey the component to the pin's guiding range through visual guidance, and the component can be accurately assembled in place through pin guidance and force feedback control. The force control threshold was determined to be 30 N. This technology can meet the engineering implementation requirements for satellite component assembly.

Ultra-precision slow tool servo (STS) machining can directly generate high-precision continuous and discontinuous freeform surfaces. However, in the fabrication of a micro-lens array (MLA) by STS, a single lens achieves different levels of quality. In addition, the low quality of a single lens may induce the failure of the entire functional component. To study the factors affecting the machining precision of an MLA on a spherical surface in STS, an experimental investigation was conducted. Specifically, the effects of the geometries of the base surface and position of a single lens on the machining precision of an MLA were examined. In the experiments, an MLA was machined by STS into three spherical surfaces having different radii, and Bruker GT-X was used to measure the base surface and micro lenses. The effects of the base surface and lens positions on the surface roughness and form error of a single lens were studied. Experimental results show that the position of the single lens changes the surface topography, surface roughness, and form accuracy of the lenses on the same base surface. In addition, the geometrical information of the base surface changes the machining precision. When the radii of base surfaces are increased from 50 to 150 mm, the surface roughness of the outer circle lens decreases from 75.78 to 69.08 nm (Ra). Therefore, considering the effect of the base surface and lens position on machining precision in the ultra-precision STS machining of MLAs is necessary. This may contribute to improving the precision consistency of MLAs and ensuring proper component function.

A two-stage lever-type micro-force generator was designed to address the challenges in the calibration system of force sensor to provide accurate micro loads. Initially, the working principle of the micro-force generator was introduced based on the performance comparison among general flexure hinges. Next, the force and energy transmissions were analyzed and a theoretical calculation method to evaluate the minification ratio K was deduced by taking into consideration the deformation of the lever and the offset of the flexure hinge's rotation. To accomplish the aim of achieving a certain minification ratio, the optimization design of the micro-force generator was proposed. Moreover, the response characteristics under different input forces were obtained by performing finite element simulation. Subsequently, a test platform was fabricated to measure the power performance of the micro-force generator. The results show that the largest error between the finite element analysis (FEA) and the theoretical analysis result is 5.501%, whereas that between the experimental result and the theoretical analysis result is 7.391%, the linearity is 2.89%, and loading range of up to 500 μN is reached. The results also indicate that the minification ratio K meets the design requirements and verify the validity of applying the optimization method to design two-stage lever-type micro-force generator and improve the accuracy of micro-Newton loads.

In order to overcome the restrictions of traditional hand-eye methods for determining hand-eye correspondence and robot-world orientation with a calibration reference, an improved hand-eye calibration approach without a calibration reference is proposed based on second-order cone programming. A relevant experimental system is established for its validation. First, a structure-from-motion approach is used to recover the camera motion matrix up to scaling. Then, the rotation and translation matrix in the calibration equation is parameterized by dual quaternion theory. Finally, the second-order cone programming method is used to simultaneously determine the optimal solution for the scale factor of the camera motion matrix, the robot-world calibration and the hand-eye calibration. Both the simulation and experimental results indicate that, for the calibration precision, the relative error of rotation is 3.998% and the relative error of translation is 0.117% in the absence of a calibration reference as a benchmark. Compared with other calibration methods, the proposed method can effectively improve the accuracy of robot-world calibration and hand-eye calibration without a reference, and extend the range of applications of the hand-eye calibration method.

An auto-control system was designed and realized for the closed-loop operation of a strontium atomic clock. This closed-loop operation locked an ultra-stable laser to a frequency obtained by averaging two peaks of the spin-polarized spectrum of the hyperfine energy structure of an isotope of 87Sr. Firstly, the requirements for automatic control of the closed-loop operation of the 87Sr atomic clock were specified, including the controlling signals and their time sequences during laser cooling and trapping of the atoms, the detection of the clock transition spectrum, and the closed-loop operation. Secondly, these specified requirements lead to the design of the physical systems. Finally, the auto-control program was designed using LabVIEW, and data acquisition hardware from National Instruments. The measurements of the frequency stability demonstrated that the in-loop frequency instability is approximately 5×10-15/τ1/2, and the frequency instability for 3 000 seconds of sampling time is 5.7×10-17. These results demonstrate that the designed auto-control system meets the requirements of a strontium atomic clock for closed-loop operation.

In the detection process for the inner-surface of deep-hole parts, the correction of image geometric distortion caused by the characteristics of a curved plane has proven to be difficult. In this study, a geometric correction algorithm for structured light stripe imaging was proposed. Firstly, the deep-hole inner-surface model (DIM) obtained via undifferentiated modeling was established. Then, the geometric position correspondence relationship between the DIM and the deep-hole inner-surface plane model was established based on discrete mapping. Finally, the geometric distortion of the structured light image was corrected based on the mapping relationship. The test results indicated that the proposed algorithm can effectively improve the correction accuracy of geometric distortion to the sub-pixel level without considering the edge of the image. The maximum distance (i.e., distance deviation) between the corresponding stripes caused by the inconsistency of the slope is the distance deviation, which was 0.135 mm less than 1.5 pixels.

Scene matching requires higher matching speed and memory usage. In order to improve the running speed of the normalized cross correlation algorithm and reduce its memory occupancy rate，this paper focus on researching the steps of fast calculating sub-image's energy. After detailed analysis, the integral graph method has the advantages of flexible and rapid, but the defect is that it needs to spend a lot of memory at the same time, while it is not suitable for the embedded system. Therefore, a fast recurrence method was proposed. In this method, the energy of adjacent pixel values is used to continuously recursive compute. It is not necessary to allocate space for all image energy as the integral image method in the calculation process. Only one row of space can be reserved for the entire energy calculation process in fast recurrence method, which greatly saves the memory usage. The fast recurrence method has the equivalent calculation speed with the integral image method, and the time consuming is only 1/2 of the traditional normalization cross correlation algorithm. In the memory occupancy rate, the fast recurrence method is less than 1/3 of the integral image method, and the larger the size of the real-time graph, the less memory occupied by the fast delivery method. In the normalized cross correlation algorithm, the classical integral graph method and the fast recursive method proposed in this paper are used to calculate the energy of the sub-image's energy, which are both faster than the traditional NCC algorithm. The two algorithms have their advantages. The classical integration image method is fast and flexible, which is suitable for the application scene with high speed requirements, but the memory occupancy rate is not very high. The fast recursive method is fast and saves memory, and is more suitable for the application of embedded systems.

In visual SLAM problems, the ORB feature has drawn much attention because of its high efficiency and stability. To address problems such as the low accuracy of image point measurements and the obvious phenomenon of feature aggregation during ORB feature extraction, a uniform distributed subpixel ORB feature extraction method suitable for high-precision SLAM was proposed. In this study, the principle of precise feature positioning was first analyzed, the error equation was then reasonably simplified, and a weight function calculation method based on template window distance was finally adopted, all of which significantly reduce the algorithm's computational cost. A quadtree-based uniform distribution solution was designed in which the image plane space is segmented with only a limited number of iterations. Features with optimal response are then exported. Experiments show that the additional computational burden of feature extraction for our method is less than 2.5 ms. The measurement accuracy of ORB features is 0.84 and 0.62 pixels on the TUM and KITTI datasets, respectively, reaching the subpixel level. Our method can thus reduce the initial value of errors and increase the efficiency of bundle adjustment. The problem of feature aggregation is effectively solved based on the condition of satisfying the overall distribution of features, which is beneficial to the robust and accurate solution of subsequent problems.

To improve the accuracy of action recognition based on the human skeleton, an action recognition method based on geometric features and a recurrent temporal attention network was proposed. First, a vectorized form of the rotation matrix was defined to describe the relative geometric relationship between body parts. The vectorized form was fused with joint coordinates and joint distances to represent a skeleton in a video. A temporal attention method was then introduced. By considering the weighted average of the previous frame, a multi-layer perceptron was used to learn the weight of the current frame. Finally, the product of the feature vector and corresponding weight was propagated through three layers of long short-term memory to predict the class label. The experimental results show that the recognition accuracy of the proposed algorithm was superior to that of existing algorithms. Specifically, experiments with the MSR-Action3D and UWA3D Multiview Activity II datasets achieved 96.93 and 80.50% accuracy, respectively.

To enhance reconstruction performance in fluorescence molecular tomography, a joint-norm and a Laplacian manifold regularization model that combined both sparsity and spatial aggregation information was utilized for light source reconstruction. In this report, sparse reconstruction by separable approximation (SpaRSA) was developed to investigate the joint model (SpaRSA-resolved Laplacian manifold regularization model, SpaRSALM). To improve the convergence speed of the SpaRSALM algorithm, a warm-start strategy was applied for light source reconstruction. The experimental results show that the SpaRSALM algorithm solved the joint model problem and improved the contrast to noise ratio (CNR) from 6.45 to 9.18 compared to using the SpaRSA algorithm to solve for the -norm regularization model. In addition, the reconstruction of the SpaRSALM algorithm using the warm-start strategy (compared to without the warm-start strategy) required 50.10 s (as opposed to 101.84 s). The accuracy and speed of light source reconstruction were significantly improved, and better reconstruction results were achieved using the presented method.

A design method for a low-light level camera driving system based on an autonomous underwater vehicle was proposed. First, the noises of an electron multiplying charge-coupled device camera was analyzed, and the principle and method of designing conventional driving circuits was proposed taking into consideration the relationship between dark current noise and clock-induced charge noise. Next, the issue of power consumption in totem-pole circuits applied to electron multiplying drivers was discussed, and a power optimization plan was presented. A high-precision system clock was used for fine adjustments of the driving signal phase and pulse width and solving the problem of amplitude overlap rate insufficiency. Finally, the low-light level camera structure and experimental results were presented. The experiments indicate that the system-generated conventional driving signal frequency is 10 MHz. The serial transfer clock amplitude overlap rate and parallel transfer clock amplitude overlap rate are better than 50% and 90%, respectively. The phase adjustment accuracy of the driving signal and pulse width adjustment accuracy are 18° and 5 ns, respectively. The driving signal is stable and smooth, the electron multiplying driving signal is highly adjustable, and the power consumption is lowered by 7.2%. The parameters of noise, size, and power consumption were considered in this design. Thus, it can be widely used in underwater low-light level imaging as well as conventional charge-coupled devices.