The application of deep learning technology in the field of computer vision is becoming more and more extensive, and the target tracking technology based on deep learning is an important research topic in the field of computer vision. In this paper, firstly, the development process of target tracking technology at home and abroad is reviewed, and some typical target tracking methods based on deep learning are introduced in detail and compared; then, several typical data sets in the field of target tracking are summarized; finally, the research progress of the target tracking technology are summarized, and the future development directions are prospected.

The latest developments of representative key technologies of low noise CMOS image sensors are summarized. The active pixel structure and the readout circuit architecture of image sensor are introduced from such aspects as the architectures and module design of CMOS image sensors. The readout circuit with inpixel source follower and its noise equivalent model are analyzed. The key technologies of low noise CMOS image sensor are introduced in detail, including referenceshared inpixel differential amplifier technique, correlated multiple sampling technique and inpixel chopping technique, and the circuit implementations of these techniques are introduced as well.

The research progresses of the techniques for measuring the tip clearance of aeroengine are introduced, and their principles, characteristics and specific applications are analyzed, and the development trends are pointed out. Based on this research, a new method for online monitoring of highspeed blades of aeroengines is proposed, and a diaphragmtype hightemperature optical fiber FP acoustic emission sensing measurement system is introduced to achieve aeroengine blade fault monitoring.

The CMOS image sensor integrated with high frame rate and high sensitivity pin photodiode array was designed. Compared with pn photodiode, pin photodiode presents the advantages of small junction capacitance and high quantum efficiency. In order to promote the signaltonoise ratio, the sensor implements a new correlated double sampling (CDS) circuit for eliminating KTC noise while working in the integration during reading mode. In this paper, linear CMOS Image sensor based on CTIA pixel circuit was implemented in 0.35μm PINCMOS process, and the photoelectric response of the device was tested and evaluated. The results show that sensitivity is 3000V/(lx·s), and quantum efficiency is 96% at 700nm wavelength. At a frame rate of 40kHz, the signaltonoise is 7 with the illumination down to 0.05lx, which is suitable for highspeed detection under weak light signal.

A kind of 128×2 linear mode APD focal plane detector used for 3D imaging is designed, including Sibased APD focal plane array (FPA) and readout circuit. The APD FPA with the structure of punch through n+pπp+ works at the linear multiply mode, and the pixel pitch is 150μm. The readout integrated circuit (ROIC) integrates the preamplifier, TDC, ADC and other function circuits on a single chip. The linear mode APD detector can simultaneously detect the laser pulse of 128×2 imaging array, then output the flight time and the amplitude of the laser pulse by using the LVDS driver. The test results show that the linear 3D imaging detector can obtain necessary information properly, verifying its effectiveness in imaging.

The core sensitive component of resonator fiber optic gyroscope (RFOG) is the fiber ring resonator (FRR), and it shows reflective or transmissive structure under different restraint modes. In this paper, the analysis models for the two structures are established, and the expression of FRR output characteristics under the best working conditions is derived. For the backscattered noise, which is the main noise affecting RFOG, the output characteristics of signal and backscattered noise of the RFOG being composed of two different structures of FRR under phase modulation are derived by establishing two phase modulation RFOG structures. Analyzed is the expression of the signaltonoise ratio(SNR) between the signal at the maximum slope of the resonance curve and the backscattered noise under the two structures, and it turns out that both SNR expressions for the two structures are the same, and it is negatively related to the cavity length of FRR.

In this article, a topemitting organic lightemitting device with the structure of Ag/Glass/ITO/TAPC/mCP/mCP:Firpic/TPBi/LiF/Al/Ag/Alq3 was developed. By growing a layer of Ag reflector on the back of the ITO glass substrate, the blue light emitted by the device was reflected by the reflective film to the top electrode to emit. The Alq3 light coupling layer on the top electrode effectively improves the transmittance rate of the metal cathode and reduces the microcavity effect of the device. The experimental results show that when the thickness of the optical coupling layer is 30nm, blue TEOLEDs with a maximum current efficiency of 8.91cd/A and a maximum brightness of 5758cd/m2 were obtained. Meanwhile, under a voltage of 10V, the CIE coordinates are (0.157,0.320), which only drift (0.002,0.010) when the brightness changes in the range of 1~5000cd/m2, showing good color stability.

A biconical MachZehnder interferometer based on photonic crystal fiber is designed, and its refractive index (RI) sensing characteristics are studied. The sensor performance was simulated by FDTD Solutions, and the results show that the transmission spectrum is redshifted with the increasing RI and the sensitivity is 95.906nm/RIU. An experimental platform was built to test the sensor under different concentrations of glycerin solution, and the response relationship between its transmission spectrum and external refractive index was studied. The experimental results show that with RI changing from 1.3222 to 1.3538, the sensitivity is 121.95nm/RIU@1550nm. The RI sensor with the characteristics of high RI sensitivity, easy fabrication and lowcost is suitable for applications in the biochemical and physical sensing fields.

Carbon electrode has the advantages of lowcost, convenient printing and the ability to isolate water and oxygen. Therefore, it is appealing to use carbon as the electrode material to achieve lowcost and highly stable perovskite solar cells (PSCs). However, conventional carbonbased perovskite solar cells (CPSCs) without hole transport layers is still facing the problems of low hole extraction rate, electron reverse transfer, and undesired recombination at the perovskite/carbon interface. In this paper, poly (3hexylthiophene) (P3HT) is applied as the hole transport layer of the device, thus the photovoltaic performance of solar cells with an architecture of ITO/SnO2/MAPbI3/P3HT/Carbon is significantly improved, achieving a power conversion efficiency (PCE) of 13.37%, which is 2.21% higher than that of the device without P3HT. Moreover, in nitrogen environment, the PCE of the device remains 87% of its initial value with continuous illumination for 1000h. In contrast, the PCE of the device without P3HT remains only 60% after illumination for 500h.

Based on the waveguide structure of metaldielectricmetal (MIM), a kind of plasma waveguide filter with doublering and fourring resonators on both sides was proposed, and its transmission characteristics were calculated by electromagnetic simulations. The physical mechanism of the peak and valley generation in transmission spectrum was analyzed by electromagnetic field distributions. The results show that the structure can realize the plasmon induced transparency (PIT). By adjusting the effective radius of the resonator, the position, bandwidth and slow light effect of the PIT transparent window can be adjusted. The results show that the signal delay of 0.148 and 0.358ps can be realized in the PIT window of double ring and four ring resonant structures, respectively. This property will have potential applications in the design of tunable filter devices, optical memory devices and integrated photon devices.

In view of the particularity of the structure of the secondary mirror base in the space environment, it is necessary to monitor its temperature state, thus a FBG temperature sensor with metal base is designed for the application of satellite secondary mirror seat. By changing the distance of fixed points, combining the principle of optical fiber temperature sensing with the strain state of bent gate region, a substrate with small structure and light weight is designed, and also it can realize multiposition temperature measurement simultaneously. Experimental results show that the linearity of the temperature sensor reaches 0.999 in three bands, and the sensitivity is 10.78, 10.81 and 10.36pm/℃ respectively for the three bands. The center wavelength of the sensor fluctuates within ±3pm at the same temperature, which is less affected by external stress and has good repeatability.

Refurbished electronic components present serious potential quality problems, which will pose serious threats to the quality and safety of space equipments. In order to ensure the quality and reliability of installed electronic components, a nondestructive measurement method is proposed for identifying refurbished components based on their characteristics. Firstly, the measurement methods and principles are described: the roughness of both the top and the bottom surfaces of the device is measured quantitatively by optical interferometry, and then the difference between the two roughness values will be used to judge whether the surface of the device was refurbished or not. Lastly, the surface roughness of the normal and refurbished components is compared and analyzed, and the uncertainty evaluation is carried out. It is shown that this method can be used to identify refurbished components efficiently and accurately. It provides a new measurement method for improving the quality of aerospace electronic components.

Front metallization is one of the most important steps during the process of fabricating singlecrystalline silicon (scSi) solar cells. The quality is proven to be closely relative to electrical properties of solar cells. According to the investigation of the influence of grid treatment techniques on the width of grids, it is found the gird will collapse to both sides in the sintering process, which might increase the shading coefficient of electrodes. With the help of characterization, the mechanism of this phenomenon was analyzed. Whats more, the effects of the slurry types, halftone opening widths, halftone patterns and the peak temperature of sintering on electrical properties were investigated. The organic content of slurry is found to affect the stability of grids in the sintering process, and the suitable opening width or design of halftone patterns can efficiently reduce the shading areas and keep the grid height consistently. On this basis, scSi PERC solar cells with a high average conversion efficiency of 22.54% were fabricated. It could be expected that through optimization of slurry and halftone pattern, improved electrical properties and higher efficiencies of scSi PERC solar cells will be realized.

In order to improve the shape reconstruction accuracy of fiber Bragg grate (FBG) shape sensor and reduce the reconstruction position error of the grating points, a FBG sensing element structure was designed in this paper. By establishing the strain transfer mechanics model of the sensing element, the expression of the average strain transfer rate was derived. And by combining with the finite element simulation model of the sensing element, the influence of the relevant structural parameters on the strain transfer rate of the sensing element was analyzed. The finite element simulation and theoretical calculation results were compared to verify the validity of the theoretical model. Furthermore, the effect of strain transfer rate on the shape reconstruction accuracy of the sensing system was analyzed, and it was concluded that when the strain transfer rate of the grating points remains above 90%, the reconstruction position error of the grating points will remain within 0.08mm. The study shows that by reasonably controlling the influencing parameters, the strain transmission efficiency of the sensing element can be effectively improved, and the position error of the shape reconstruction of the sensing system can be reduced, thus the positioning accuracy of the shape reconstruction of the optical fiber shape sensor will be improved.

A dualmode largemodearea multicore optical fiber is proposed in this paper, and the electromagnetic field model of the proposed fiber is constructed and solved by the finite element method. Both the mode and the bending characteristics of optical fiber are studied based on the model. The influence of the three parameters of core pitch, core radius and refractive index difference between core and cladding on the mode characteristics and effective mode field area of the fiber is analyzed systematically. The proposed multicore fiber can realize strict dualmode transmission by using air holes and reducing the core pitch, core radius and refractive index difference between core and cladding. When the optical fiber maintains dual mode transmission, the area of the mode field of the fundamental mode will increase with the increasing core pitch and decreasing fiber cores radius and the refractive index difference between core and cladding. If the dualmode transmission is approximatively met, the mode field area of the fundamental modes of the straight fiber can reach 3155μm2 by adjusting the structural parameters.

In order to improve the sensitivity of the edgelapping (Dshaped) plastic optical fiber (POF) evanescent wave sensor, the optical fiber grinding paper with different particle diameters was used to grind the uncoated POFs to prepare Dshaped POF evanescent wave sensors with different diameters and different surface roughness. The surface morphology of the Dshaped region was checked, and the effects of the diameter and surface roughness of the Dshaped region on the light transmission and sensitivity of the sensors were investigated experimentally. Experimental results indicate that the diameter and surface roughness of the Dshaped sensitive region have a significant effect on the light transmission and sensitivity of the sensor. The sensitivity increases first and then decreases as the diameter of the Dshaped region decreases, and when the diameter is 1200μm, the sensors response sensitivity to glucose solution reaches a maximum of -0.0032(mg/L)-1; when the Dshaped region is polished with the fiber optic paper with the particle size of 9μm, the sensors sensitivity will be further improved to -0.0045(mg/L)-1, which is 11.25 times of that of POF sensor without grinding.

Multilayer films with the structures of ITO/SiO2/ITO, ITO/Ti2O3/ITO and ITO/MgF2/ITO were coated on the K9 glass substrate by electron beam evaporation coating method. The square resistance on the surface of the films was measured by fourprobe square resister and the surface microstructure of the samples was observed by atomic force microscope (AFM). The results show that when the roughness of ITO film is large and the physical thickness of the medium film is less than 100nm, each layer of ITO films is connected through some bump structures like mountains, resulting in that the measured square resistance of the surface of the sample film is almost the same as the square resistance in parallel of each layer of ITO films. This indicates that when ITO film roughness is large and medium film thickness is small, it will cause the resistance of each ITO film to be in parallel. The 450~1200nm ultrawide spectrum transparent conductive film was designed and prepared by using the square resistance parallel effect of multilayer ITO films. The surface square resistance of the test sample was measured with a fourprobe square resistor and the spectrum transmittance of the sample was measured with a spectrophotometer. The results show that the multilayer transparent conductive film prepared by the square parallel resistance effect of ITO films has a higher spectral transmittance than the single layer ITO film under the same surface square resistance.

The pore structure, size and type of electrode material directly affect its electrochemical properties. In this paper, threedimensional porous graphene electrode materials were prepared by a twostep method of hydrothermal reaction and nitric acid steam treatment. The threedimensional porous graphene was characterized by scanning electron microscope, transmission electron microscope, Raman spectrum and Xray diffraction. And the electrochemical properties of the porous graphene were tested by the threeelectrode method and electrochemical methods such as cyclic voltammetry, constant current chargedischarge and electrochemical impedance. The results show that the prepared 3D porous graphene has a combined structure of micropores and nanopores, and their synergistic effect leads to excellent electrochemical properties of 3D porous graphene, realizing a specific capacitance up to 191.5F/g at the current density of 1A/g.

Taking the pulsating heat pipe with corrugated configuration and the microgrooved flat heat pipe as the research objects, a threedimensional unsteady mathematical model was constructed based on the Mixture model, and the reliability of the model was verified. This mathematical model was used to compare the thermal resistance, average wall temperature and the uniformity of the evaporation section wall temperature of the two types of micro heat pipes under the same heat dissipation space and heat flux. The results show that, compared with the microgrooved flat heat pipe, the pulsating heat pipe with corrugated configuration presents lower thermal resistance and better heat transfer performance; the steadystate average wall temperature of the evaporation section of the pulsating heat pipe with corrugated configuration is lower, and this advantage becomes more obvious as the heat flux increases; the pulsating heat pipe with corrugated configuration has better wall temperature uniformity in the evaporation section on the spatial scale, and this advantage is more prominent in the case of high heat flux, but this uniformity changes relatively drastically on the time scale.

Aiming at the problem of approximate symmetry of the feature points in the pin matching process of the electrical connector and the largeangle rotation transformation between different images, it proposes a feature point matching algorithm based on the invariant of the sixpoint characteristic number. In this paper, the feature points were divided into two parts of convex hull and interior point. Then the invariance of the order of the points on the convex hull in the projective transformation was used to achieve the convex hull matching, and the similarity of interior point feature vectors based on convex hull feature points was used to achieve the interior point matching. The experimental results prove that the proposed algorithm can achieve the matching of the pin feature points well with certain robustness.

Phasebased stereo matching (PSM) is a vital step in binocular structured light measurement, but the traditional PSM method is difficult to strike a balance between speed and accuracy due to the greatly increased storage space when processing highresolution images. In this paper, an image pyramid model is introduced and a phasebased stereo matching algorithm based on multiscale analysis is proposed. With a layered matching strategy, the preprocessed left and right absolute phase maps are downsampled to generate an image pyramid and highresolution disparity maps are predicted by upsampling lowresolution disparity maps to narrow the disparity range. Experimental results indicate that the proposed algorithm can greatly improve the phasebased stereo matching speed while ensuring the accuracy and can achieve fast and accurate stereo matching of highresolution images.

Aiming at the problems of large amount of calculation and slow speed in the process of image registration for traditional Harris corner detection algorithm, in this paper, a fast prescreening Harris corner detection algorithm was proposed. Firstly, a large number of nonfeature points were quickly eliminated with the FAST algorithm, and then the clustering of FAST corners was solved by suppressing the radius. Secondly, Harris corners in the neighborhood of FAST corners were selected, and finally BruteForce matching was used to realize accurate matching. Experimental results show that the proposed algorithm not only improves the corner detection speed but also reduces the number of redundant corners, and it effectively improves the registration speed and accuracy, realizing a good registration effect.

A novel construction method of ratecompatible lowdensity paritycheck(RCLDPC) codes based on the matrix extension and combined with the properties of Fibonacci sequences is proposed in order to reduce the hardware resource in satellite laser communication systems. The RCLDPC code with girth6 and quasicyclic characteristics can be constructed by this construction method, which can save the storage elements and reduce the computational complexity. Its hardware implementation was more easily realized and the RCLDPC code is more suitable for satellite laser communication systems. The simulation results show that the RCLDPC code constructed by this method can realize a better decoding performance in a wider coderate range. Under the same parameters, the net coding gain of the constructed RCLDPC code is more than those of other codes with the same coderate and codelength at the bit error rate (BER) of 10-6.

The amplitude modulation and phase modulation are the key techniques in laser communication systems, but there is a limitation in satellite network with intersatellite links as single laser terminal can only have a specific modulation format. In this paper, an optical modulation technique which is compatible with multimodulation formats and bit rates based on digital processing was described. A LiNbO3 quadrature modulator combined with closedloop bias control algorithm was applied to implement amplitude modulation (OOK modulation) , phase modulation (BPSK modulation) and multilevel phase modulation (QPSK modulation) in a signal modulator. The modulation bit rate can be tuned from 625Mb/s to 5Gb/s according to different modulation formats of OOK, BPSK and QPSK. The transmission EVM is less than 9% with coding and scrambling functions and the inducted loss for the receiving sensitivity is less than 1dB. This modulator has been verified in all environment experiments such as vibration test, thermal vacuum circle test and radiation test and could be used in various laser communication terminals. The modulator realizes the function of analog modulation, so it is expected to be used as a transparent transponder for microwave photonic signal.

In order to measure the shape of a minimally invasive surgical soft manipulator when it is flexing and stretching, a shape sensing method based on spirally arranged fiber grating is proposed. Unlike the traditional linear layout method, the spiral layout method can prevent the fiber grating from breaking and degumming when the soft manipulator flexes. The sensing principal of spirally arranged fiber grating and the reconstruction algorithm of bending deformation are analyzed theoretically. The sample of fiber grating placed on the surface of the soft manipulator along the spiral is prepared, and the experimental system for bending deformation of the soft manipulator is established. The variation of the center wavelength drift and bending angle of spirally arranged fiber grating under different bending conditions is analyzed experimentally, and the deformation shape of the soft manipulator is reconstructed. The experimental results show that the maximum sensitivity of the spirally arranged fiber grating in different positions of the soft manipulator is 7.1pm/(°), and the maximum error between the actual measured value of the bending angle of the soft manipulator and the theoretical reconstruction value is less than 4.29. The helical optical fiber sensing method can be used for the shape sensing and reconstruction of soft manipulators.

When Camshift algorithm is applied to target tracking of NAO robot, the tracking fails if the target is interfered by similar color background or blocked by objects. A target tracking method based on ORB feature detection and Kalman filter is proposed. Firstly, it is to detect the target ORB signature point and initialize the search window. Then Kalman filter is used as the prediction mechanism of target motion state to initialize Camshift algorithm with the predicted position. The Bhattacharyya distance is used to determine the convergence of the trace window. If background interference occurs, the ORB algorithm is used to match feature points between Kalman prediction area and target model in the current frame, and the position of the target in the video frame is redetected. According to Kalman filter, the possible position of the target after it is blocked by the object is updated. The experimental results show that the improved algorithm can continuously and stably track moving targets under the complex environment of similar color background interference and target occlusion.

In order to solve the problem of linear demodulation for phasemodulated microwave photonic link, a theoretical study of highlinear phase demodulation with optical phaselocked loop (OPLL) is carried out. A detailed link model is established, and taking full account of the influence of noise and nonlinear distortion, the variation trend of key parameters of the link was derived. And also, the key parameters of the link were simulated based on the parameters of actual link devices. Simulation results show that based on the OPLL, the dynamic range of the link can reach 127dB·Hz2/3 at 200MHz RF, which is nearly 20dB higher than that obtained by the traditional intensity modulation method, providing a reference for applications of largedynamic microwave photonic link.