An embedded photoelectric search aiming and indication system suitable for stealth airborne platforms was developed, which has the functions of wide-area reconnaissance, staring surveillance, laser ranging and laser irradiation. The wide-area scanning control technology for the optical image motion compensation of geographical stability combined with fast-steering mirrors was deduced and verified, and the algorithm verification was carried out in the laboratory and outfield based on the prototype. The test results show that under the premise of clear imaging, the proposed system realizes the two-dimensional seamless wide-area scanning search function, which can splicing and display the scanned video images in real time and meets the functional requirements of information warfare for target reconnaissance, surveillance and indication.

To realize high-efficiency countermeasures and stealth for future remote terahertz radar, a wide-band terahertz absorbing structure with graphene was designed for typical operating frequencies of terahertz radar. The basic absorbing structure unit of wide-band absorbing structure was the surface metal layer/graphene layer/dielectric layer/bottom metal layer, and the dual-scale four separation layers were designed and optimized to certain the structure parameters of each layer using the genetic algorithm. Simulation results show that the absorption efficiency of wide-band absorbing structure is better than 80% in the frequency range of 0.138 THz~2 THz, 97.46% in the frequency range of 0.157 THz~2 THz, and 92.27% at the typical terahertz radar operating frequency, which satisfy the demands of countermeasures and stealth for terahertz radars.

In recent years, the endoscope is widely used in defect detection of small-size parts in complex environment. An industrial endoscope optical system was designed for aero-engine blade detection. The basic structure of the lens adopted secondary imaging, and the objective lens adopted asymmetrical inverse distance structure, which beamed the large field of view light into the small caliber cavity. The adaptor lens was adapted to the high speed camera in which the diagonal length was 42 mm, and it could magnify the intermediate image 21 times. The system was based on Zemax design software for system optimization, tolerance analysis and image quality evaluation, and finally the system realized the optical properties of large field of view (120°), small caliber (3 mm) and high temperature resistance (25 ℃～180 ℃). Due to the higher requirements of the field of view, caliber and adaptor magnification, the aspherical surface was reasonably introduced to improve the imaging quality of the system. The entrance pupil diameter was increased to 0.5 mm, the system spatial cut-off frequency was at 17 lp·mm?1, the full-field modulation transfer function value was greater than 0.28, and the maximum distortion value is less than 21.2%.

To meet the requirements of short-range and large-screen projection system in domestic and foreign markets, it is necessary to reduce the processing difficulty and cost of projection lens. In this case, an ultra-short focal projection lens was designed. The catadioptric structure was selected, the principle of the relationship between object and image was used, the mirror surface type parameters were calculated by combining the object field of view, and finally the mirror surface type was obtained. This method was convenient to correct the distortion of large field of view and counteract the astigmatism and field curvature of refraction lens group. The whole optical system consists of 15 spherical lenses and one piece of aspheric mirror, with a total length of 286 mm, projection distance of 500 mm, projection screen size of 254 cm (100 inch), and projection ratio of 0.22. The relative illumination is greater than 85%, the MTF is greater than 0.5 at 0.4 lp/mm, the horizontal TV distortion is less than 0.55%, and the vertical TV distortion is less than 0.65%. Many design indexes of the system are better than that of the existing products on the market, and its coaxial spherical surface design is easy to process and adjust, which can effectively reduce the production cost.

To meet the needs of further miniaturization and cost reduction for existing immersive flying simulator, a spherical zone projection lens for small and medium-sized flight simulator was designed. The key parameters of each component (spherical zone, projector, projection lens) in the projection scheme was determined. The designed lens has a field of 172° and inverted-telephoto ratio of 6:1, put a projector with the designed lens on the screen, and the whole screen would be fully projected. The construction method of optical system with large field of view and inverted-telephoto ratio was provided, the distortion, field curvature and relative illumination in large field nonsimilar imaging systems were analyzed and discussed, and the effective method of improving the illumination uniformity was also presented. The proposed lens meets the high-quality projection requirements of 4 k resolution with a low-cost structure of 8-piece lens, and its MTF values of all the fields are higher than 0.4 at the nyquist frequency of 185 lp/mm, which has good workability.

Aiming at the problem of lack of research on optimizing thermal condition of ultraviolet spectrometers, a Rowland circle spectrometer with a detection range of 200 nm~450 nm and a full-band resolution of no less than 0.2 nm was designed, and optimized the thermal conditions of its optical chamber by coupling optical, thermal and structural simulations. The thermal simulation results showed that the temperature and temperature difference of the spectrometer base was increased with time in the absence of heating and inlet wind speed, and it was difficult to achieve thermal balance. The inlet wind speed of optical chamber was optimized, and it was found that when that was 0.8 m/s, the overall temperature was dropped to 36.103 ℃~39.859 ℃. Based on the calculation of thermal deformation between optical devices, the intercept of total thermal deformation of several optical devices was 0.203 mm. After refining the heating method, the top-layer heating was found to be the best way, the overall temperature was dropped to 34.241 ℃~36.139 ℃, and the intercept of total thermal deformation was reduced to 0.122 mm. The optical simulation results show that the optimized Rowland circle spectrometer can still clearly distinguish the two beams with a wavelength difference of 0.2 nm after thermal deformation.

For the conventional color separation photography technology was difficult to obtain the clear image information and rich texture characteristics, a trace color separation photography intelligent system based on multi-band light source was designed. The multi-band light source module was designed by STM32, LED, additive color effect and complementary color principle to solve the problem of selecting monochromatic light corresponding to different trace backgrounds. The light source module that could be moved up and down, left and right was designed by hybrid stepper motor, DEV8825 motor driver module and STM32 to solve the problem of finding the best light distribution position. Through the improved FS-SIFT registration algorithm and the introduction of the second-generation Curvelet fusion algorithm, a combined image processing system was designed to solve the problem of unsynchronized trace extraction and image processing in conventional photography. The experimental results show that the brightness contrast between the background and the trace image photographed by the system is larger and the quality is better. After the improved FS-SIFT algorithm registration and the second-generation Curvelet algorithm fusion processing, the average standard deviation of the image is increased by 1.25 times, the average information entropy is increased by 1.70 times, the average gradient is increased by 1.46 times, and the average amount of information and texture features of the obtained image are richer.

Due to the lack of information and the symmetrical imaging of single-station imaging, the ambiguity of the optical single-station pose processing is a common and difficult problem. Aiming at the ambiguous solution of optical single-station pose processing, the corresponding relationship between the target subject straight-line vector and the pose angle was taken as the breakthrough point, and the object-image mapping relationship of the target subject straight-line vector was taken as the basis of single-station pose processing. Defined the partial ambiguous error solution from the single-station pose processing results, and then the ambiguous solution was modified according to the key steps of single-station pose solution. The effects of processing results before and after ambiguous correction were significant compared with that of the conventional intersection processing results. In the described scenario, the error magnitude is reduced by at least 20%. This method provides a reliable theoretical support for obtaining the correct solution of optical single-station pose processing.

The edge contour features of infrared face images are of great value for applications related to infrared face detection and recognition. Aiming at the problem of false edges in the edge contour extraction of infrared face images, an edge contour extraction of infrared face image based on improved Canny algorithm was proposed. Firstly, by introducing dynamic threshold constraint factor to the guided filtering algorithm to replace the Gaussian filtering in the original algorithm, the disadvantages of uneven filtering processing and the loss of weak edge features in the infrared face image were solved. Then, the non-maximum suppression was improved, and four gradient directions were added on the basis of the original gradient direction, which made the interpolation of non-maximum suppression more precise than the original algorithm. Finally, the OTSU algorithm was improved by constructing a gray-gradient mapping function to determine the optimal threshold value, which solved the limitation of the original algorithm to determine the threshold value by human experience. The experimental results show that, compared with the original Canny algorithm filtering processing, the performance of signal-to-noise ratio of the filtered image from edge contour extraction of infrared face images based on improved Canny algorithm is improved by 34.40%, and the performance of structural similarity is improved by 21.66%. Finally, the quality coefficient value of the experiment of infrared face edge contour extraction is higher than that of other methods in the comparison experiment, which proves that the improved algorithm has superiority for the edge contour extraction of infrared face image.

Aiming at the phenomenon that the accuracy of convolutional neural network is easy to be saturated in gait recognition and the problem of low fitting efficiency of vision transformer (ViT) to gait data set, an idea to construct a symmetrical dual attention mechanism model was proposed to retain the time order of walking posture, and fit the gait image blocks with several independent feature subspaces. At the same time, the symmetrical architecture was adopted to enhance the role of attention module in fitting gait features, and the heterogeneous transfer learning was used to further improve the efficiency of feature fitting. The model was applied to CASIA C infrared human body gait database of Chinese Academy of Sciences for many simulation experiments, and the average recognition accuracy was 96.8%. The results show that the proposed model is superior to the traditional ViT model and CNN comparison model in stability, data fitting speed and recognition accuracy.

Gray codes are widely used in structured-light projection three-dimensional (3D) imaging because of good robustness and noise immunity. During the process of 3D measurement, due to the influence of equipment and other environmental noises, the gray-code decoding fringe-order edge and the truncated phase edge usually cannot be in an ideal alignment state, which causes the unwrapped phase to jump. In order to better avoid the level hopping error and make the error tolerance width of the edge hopping region larger, an inter-partition phase unwrapping method based on Gray codes was proposed. A Gray code image was added on the basis of the complementary Gray code, the additional code words were obtained by decoding all the Gray codes, and two auxiliary fringe orders were obtained by performing fringe order mapping with different displacements on the additional code words. Using all the fringe orders, the inter-partition phase unwrapping was performed on the truncated phase. When the error in the edge hopping region was greater than half a period, the unwrapping phase without hopping could still be obtained. The experimental results show that when the width of the edge error region is less than 3/4 of the stripe period width, the error caused by the level hopping can be effectively avoided.

For the small-size defects of metal workpiece surface and the difficult segmentation of image defects due to non-uniform illumination, an improved U-net semantic segmentation network was proposed to achieve accurate image segmentation of surface defects for metal workpieces. Firstly, the convolutional block attention module (CBAM) was integrated into the U-net netwok to improve the significance of the defective targets in the image. Secondly, part of the traditional convolution in the network was replaced by depthwise over-parameterized convolution (DO-Conv) to increase the number of learnable parameters of the network. Then, the Leaky Relu function was used instead of the partial Relu function in the network to improve the feature extraction ability of the model for the negative intervals. Finally, the median filtering and non-uniform illumination compensation method were used for image preprocessing, so as to reduce the effect of non-uniform illumination on the surface defects of metal workpiece images. The results show that the improved network mean intersection over union, accuracy rate and Dice coefficient index reaches 0.833 5, 0.933 2 and 0.867 4, respectively. The improved network significantly improves the segmentation effect of surface defect images of metal workpieces.

Particles size detection is an important link in production, and the use of cameras to capture and process images is a commonly-used non-contact detection method. To meet the requirements of identification and size detection of particles, the sand particles were selected as the detection object, and a Mask R-CNN model with the improved boundary mask of particles was proposed. Combined with the classical edge detection technology, the deep learning model was used to predict the mask, and the mask with higher precision was obtained according to the results of edge segmentation. The DenseNet was used as the backbone network of the network detection to reduce the number of network parameters, and the channel attention mechanism was used to strengthen the feature extraction ability of the network. The experiments show that the improved network can improve the detection accuracy, and the combination of image processing can improve the accuracy of mask size detection, which provides a meaningful method for industrial detection of particles.

Defect detection of product surface is an important part of industrial automatic production and the accuracy is the main index to evaluate the reliability of automatic detection system. Based on the particularity of defect detection on complex texture surface and the requirements of real-time and universal detection methods, a detection method for optimizing the backbone network and using the transfer learning feature mapping to construct the complex texture surface defects was proposed. In this method, the ResNet model was optimized and the simulation data set was established for transfer learning, so as to solve the problems such as the small number of samples in the data set of complex texture surface products, the difficulty of data set making, and the difficulty of similar problems to be compatible with each other. Experimental results show that the proposed method can accurately detect the surface defects of artificial wooden plank with random complex texture, and the average accuracy can reach 99.6%. Under the existing experimental conditions, the detection time of a single artificial wooden plank is 305 ms, which can meet the real-time requirements of online detection. The research results can provide a new idea and theoretical reference for the defect detection on complex texture surface based on deep learning.

The accuracy of the flying projectile speed measurement is directly affected by the detection light screen sensitivity of the split large-target test system. In the case of small dispersion, the artificial light source in the rectangular large-area test system can be replaced by a reflective film and a laser to simplify the test system. The sensitivity of the new triangular detection light screen was analyzed. The light intensity attenuation of lasers at different propagation distances, the non-uniform distribution in space, the retro-reflection coefficient of the reflective film and the off-axis effect of the lens were taken into account. The sensitivity of projectiles of the same diameter through different positions of the light screen was analyzed by equivalent normalization at the same reference point by using numerical simulation and firing test. The simulation results show that the sensitivity is most affected by the lens off-axis effect and least by the non-uniform distribution of the laser space. The firing test was carried out in the right-angled triangle detection light screen area of 3.5 m×2 m (width×height). The results are consistent with the simulation results, the closer the lens, the higher the sensitivity, and the lower the versa. The results can provide a reference for the engineering design of a large-area detection light screen similar to the reflection type.

The optical interference absolute test technology can separate the surface figures of the reference plane as well as the tested optics, and it is an effective means to calibrate the accuracy of the interferometer. For the calibration requirements of large-aperture plane interferometer, the rotation translation method only requires one transmission flat and one reflective flat, which avoids the cost and difficulty of processing an additional flat. However, with the increase of the caliber, the weight and support make the deformation of the reflective flat larger in various states of translation and rotation, and then affect the absolute test accuracy. Therefore, it was proposed to design a lightweight calibration mirror as a reflective flat, and used the rotation translation method to realize the absolute test of the large-aperture interferometer. The Φ 1 500 mm plane interferometer was used as the test requirement, the SiC was used as the calibration mirror material, and the lightweight design was carried out with a triangular lightweight structure and a six-point back support method. The weight was controlled to only 93 kg, and the surface deformation PV value introduced by the support and gravity was 9.75 nm. It superimposed this deformed surface shape to PV value of λ/4, and performed rotation translation absolute test simulation calculation for processing surface shapes of different distributions. For the surface shapes with low degree of rotational symmetry and containing more high-frequency components, the test accuracy is λ/30, and for the smooth and symmetrical surface shapes, the test accuracy can reach to λ/50. Therefore, in order to achieve the λ/50 accuracy calibration target for the large-aperture plane interferometer, it requires that the PV value of the processed surface of the SiC calibration mirror is lower than λ/4, try to avoid high-frequency components, and have a high degree of rotational symmetry.

Aiming at the problem that the length of the temperature sensing area is smaller than the spatial resolution, which leads to inaccurate temperature measurement, a multi-point temperature correction scheme was proposed, and a device and method that could be applied to the measurement of temperature distribution in medium and small scale sections were introduced. Using the scheme based on multi-modal Gaussian fitting, the simultaneous correction of multiple inaccurate temperature points was realized, and the temperature accuracy of the system was improved. A prototype of the distributed Raman thermometry system was designed, and the measurement of temperature distribution in medium and small scale sections was realized by wrapping the fiber around the PPR tube. The experimental device improved the measurement accuracy of 2 m and 1.5 m temperature sensing areas to 1 °C, the thickness of the ice sheet of the Wanjiazhai Reservoir on the Yellow River in winter was measured to be 42.33 cm, which was 1.67 cm different from the artificial measurement results, and the measured temperature of 36 m deep river water profile fluctuated in the range of 0.32 ℃ ~ 0.90 ℃. The experimental results show that the distributed Raman thermometry system can monitor the temperature distribution of the river flow profile on a large scale in the field.

Compared with intensity detection, the polarization detection can effectively enhance the accuracy of ground object target (such as camouflaged target) recognition. A polarization properties test system was designed and built to conduct the multi-angle polarization imaging experiments on five targets, namely iron plate, glass plate, green lacquer-coated iron plate, green lacquer-coated glass plate and turf. First, the incident light source was set to completely linearly polarized light by rotating the polarizer. Then, a filter was added in front of the polarized camera lens to obtain the images of five materials in the same central band at various detection zenith angles and detection azimuth angles. Finally, the polarization degree of the obtained polarization image was calculated. The results show that the polarization preserving properties of different materials are different, and the polarization degree of the object is only related to the surface coating of the material, and has nothing to do with the internal composition. In the direction of specular reflection, the target has the largest degree of polarization. This indicates that the polarized light detection can be used as a basis for material classification, which can be applied to remote sensing detection, material evidence search and other aspects, and has certain guiding significance for the production of special camouflage materials for anti-polarization reconnaissance.

Pavement state sensor is an important tool for qualitative identification and quantitative measurement of pavement state, and its quantitative measurement performance depends on the accuracy of quantitative calibration model. In order to solve the problem of nonlinearity and nonuniform distribution of quantitative calibration data of pavement state sensor, a quantitative calibration model of pavement state sensor based on particle swarm optimization-adaptive support vector regression (PSO-ASVR) was proposed. Firstly, the adaptive preprocessing (AP) process was constructed to optimize the pre-processing of the calibration data to reduce the calibration data processing error under the influence of the nonuniform distribution of the pavement state sensor. Then, the support vector regression (SVR) algorithm based on structural risk minimization was used to fit the calibration data, and the particle swarm optimization (PSO) algorithm was used to realize the parameter optimization in the SVR to reduce the data fitting error introduced by the nonlinear calibration data of the pavement state sensor. Experiments on calibration data processing under different pavement states show that the root-mean-square error (RMSE) of the new method can be reduced by at least 63% compared with that of the traditional method, which verifies the effectiveness of the new method in improving the accuracy of the quantitative calibration model and realizes the reduction of the quantitative calibration error of the pavement state sensor.

The cavity ring-down method is currently the only method for measuring ultra-high reflectivity (reflectivity is greater than 99.9%) of optical elements. An optimal extraction method for the relationship between laser signal intensity and time in cavity ring-down method was introduced, and a test system for ultra-high reflection ratio of optical elements based on cavity ring-down method was designed. The data of cavity ring-down curve were divided into five sections by piecewise exponential fitting, the R2 (R-square) and root-mean-square error (RMSE) values corresponding to the index fitting results of each section were compared and analyzed, and the ring-down time of each index fitting was calculated. The experimental results show that the results obtained from 40%~60% of the cavity ring-down curve data are closest to the real value, and the reflectivity of the corresponding cavity mirror is 99.988 977%. Finally, by comparing with the reflectance of the cavity mirror, it is shown that this data fitting method can effectively measure the reflectivity of the cavity mirror and reduce the error caused by the experimental data.

In order to meet the requirements of helicopter rotor dynamic load test, a three-dimensional (3D) dynamic deformation measurement and visual analysis method for helicopter rotor was proposed. Firstly, according to the structure and high-speed rotation characteristics of helicopter rotor, a dynamic deformation image measurement and monitoring as well as analysis system of rotor with binocular high-definition camera as the core was designed. Then, based on the binocular stereo vision measurement theory, the principle/method of the measuring system calibration, real-time single point deformation measurement, speckle image matching, rotor surface 3D reconstruction and 3D dynamic deformation visual analysis were discussed. Finally, the simulation test of simulated and measured environment was carried out on the ground, and the test system construction, test data acquisition, processing and analysis were realized. The test results show that this method can obtain the positioning measurement accuracy with the maximum error less than 4 mm, and can well realize the 3D dynamic deformation measurement of helicopter rotor, which provides intuitive and reliable data and technical support for the flight test rotor load test data analysis.

In order to solve the problem of extreme luminance contrast in light pollution, an extreme luminance measurement method based on high dynamic range imaging technology was proposed for complex lighting environment. The 24 color standard color card and CS-2000 spectrophotometer were used to calibrate D5300 digital camera, the fitting relationship between the luminance distribution image with high dynamic range and the stimulus value of CIE 1931-XYZ surface color system was obtained, and the accurate measurement of luminance distribution in extreme luminance contrast environment was realized. Through the field measurement in urban roads, it was verified that the dynamic range of luminance measurement of extreme luminance measurement method based on high dynamic range imaging technology could reach 104, and the relative errors between measured value and standard luminance value measured by CS-2000 spectrophotometer were ?2.2% and ?2.5% respectively. It shows that this method has good applicability, which can provide metrology support for efficient and high-accuracy measurement of road lighting quality, and also provide the solution for effective prevention and control of complex light environment problems such as light pollution.

Fringe projection profilometry is widely used in many fields due to its advantages of high speed, high accuracy, and robustness to ambient lighting and surface textures. However, it is susceptible to high dynamic range (HDR) objects. Therefore, a method to quickly calculate the optimal projected brightness was proposed. By solving the response relationship between camera and projector, and combined with a proposed simple method of projection brightness determination, the required projection brightness could be obtained. Then, based on the proposed image fusion algorithm, the obtained original images under each brightness were fused to obtain high-quality fused images and achieve high-precision three-dimensional reconstruction of HDR objects. Compared with the traditional methods, there was no need to blindly project and take a large number of pictures or calculate complex homography matrices. It only needed to project a uniform white light to the measured object, and the response function between the camera and the projector could be quickly solved, so as to obtain the required projection brightness, which greatly improved the measurement speed.

In the process of preparing entangled photon pairs by spontaneous parametric conversion technology based on second-order nonlinear effects, the periodically polarized crystals doped with 5 mol%MgO:PPLN were studied. By linking the momentum conservation and energy conservation conditions in the process of optical parametric transformation with the dispersion equation of the crystal and the thermal expansion equation of the crystal polarization period with temperature, the period-tuning and temperature-tuning properties of entangled photon pairs were obtained at the five commonly used wavelength points of 355 nm, 405 nm, 532 nm, 780 nm and 1 064 nm. It was found that the crystal polarization period was too small and two pairs of entangled photon pairs were generated during the research process, and the entangled light band range generated by the interaction of each wavelength point with nonlinear crystals under a certain polarization period and temperature was summarized. When other nonlinear periodical polarized crystals were selected for experiments, the polarized periodic term in the QPM momentum conservation condition was changed, and the dispersion equation was changed according to the specific crystal used. This research scheme can be directly extended to the research of using different nonlinear crystals to generate entangled photon pairs in the communication optical band or infrared optical band, and has important research value in the field of preparing quantum light sources.

A method for designing and preparing the rectangular-wave broadband pass filter was put forward after a deep study. A rectangular-wave OD3-A broadband pass filter was designed and prepared with this method, with its center wavelength λ0=515 nm in the working range of 400 nm~1 100 nm. And the mean transmittance of the sample in its pass-band of λ0±25 nm was great than or equal to 92.7%, meanwhile the transmittance in the cut-off band λ ranging from 400 nm~475 nm and λ ranging from 555 nm~1 100 nm was less than 0.1%. The transmittance of the thin film sample was tested, and the spectrum met the requirements. The rectangular-wave broadband pass filter designed and prepared by this method overcame the disadvantages of high precision requirements of Fabry-Perot interferometer narrow-band filter film monitoring, narrow pass-band broadband, high cost, overlarge total thickness, low pass-band transmittance and poor waveform rectangularity in the traditional combination way of long and short wave cut-off films.

The finite element method was used to study the deformation distribution of the optical window of the multi-component gas in-situ laser detection system under different thermal pressing loads, and the refractive index distribution of the optical window was calculated according to the thermo-optic effect. The influence laws of the refractive index change and deformation of the optical window before and after nitrogen purging on the luminous flux and irradiance distribution reaching the receiving surface were compared and analyzed by ray tracing method. In addition, based on the laser transmission test platform of high-temperature optical window, the influence of high-temperature optical window on the detection signal was studied. The results show that the high-temperature and high-pressure environment will increase the refractive index change amplitude and deformation of the optical window, which resulting in the laser optical path deflection induced transmitted light intensity loss and probe signal fringe interference. The nitrogen purging can effectively improve the laser transmission conditions, increase the luminous flux reaching the receiving surface, optimize the irradiation distribution and improve the laser transmission quality.

In order to meet the requirement of active obstacle avoidance of microminiature unmanned aerial vehicle (UAV) in flight mission, an obstacles detection method based on monocular vision and active laser lattice projection of microminiature UAV for obstacles avoidance was proposed. The projected laser lattice patterns were collected by a monocular camera, and through the processes of image segmentation, clustering and centroid extraction, the ambiguity of the characteristic consistent laser point was quickly eliminated by the constraint of the laser line equation of the image plane. The laser points were used to detect the distribution of obstacles in the front space of the UAV. The experimental results show that the relative error of obstacles detection is within 1.5% when the baseline distance is 65 mm and the working distance is 7 m. The proposed method has high accuracy and low time complexity, and can meet the requirements for obstacles detection methods of microminiature UAV with low computing power, which provides the data support for the generation of further obstacles avoidance strategies.

In the line structured light three-dimensional measurement system, the high-precision laser stripe center-line extraction is the key to improve the measurement accuracy. Aiming at the existing laser center-line problems such as low extraction accuracy and poor retention of details, a laser center-line extraction algorithm based on normal guidance was proposed. The specific implementation steps of the algorithm were as follows: Firstly, the image was pre-processed, and the laser line was preliminarily extracted by combining edge detection and geometric center method. Then, the principal component analysis (PCA) was used to obtain its normal line, divided the angle eight neighborhood at the laser center point, and searched for the effective point set through the normal angle guidance. Finally, the gray centroid method was used to extract the sub-pixels from the point set. The experimental results show that the root mean square error of the algorithm is improved by 0.233 9 pixel compared with that of the gray centroid method, which can retain the light strip details better than the Steger algorithm and the directional template method, and can extract the light strip center more accurately and achieve the sub-pixel level accuracy.

In order to realize the application of fiber grating sensor in wearable system, a wearable fiber grating sensor demodulation system based on silicon-based photonic integrated chip was proposed. The photonic integrated chip based on Belgium iSiPP50G process was composed of 4×1 long wavelength VCSEL array, 1×8 array waveguide grating, 2×2 MMI coupler, 4×1 fiber grating coupler array, GE-on-SI waveguide photodetector, straight waveguide and curved waveguide, etc. After completing the gold wire bonding of the VCSEL light source and the optical fiber coupling package of the photonic integrated chip, a wristband demodulation circuit was designed to measure the human body temperature and heart sound signals in real time. The experimental results show that the dynamic wavelength detection range of the demodulation system is 1 540 nm~1 560 nm, the wavelength resolution is 0.08 pm, the demodulation accuracy is 5 pm, the temperature monitoring range is 35 ℃~42 ℃, and the error is ±0.1 ℃. It can detect the heart sound signal in the frequency range of 50 Hz~100 Hz, identify the first heart sound and the second heart sound, and calculate the cardiac cycle, heart rate, the first heart sound time limit, the second heart sound time limit and cardiac parameters.

A spiral-shaped plastic optical fiber (POF) surface plasmon resonance (SPR) refractive index sensor based on wavelength modulation was studied. First, the plastic optical fiber was prepared into a spiral by mechanical hot pressing and deforming method, and then a metal film with a certain thickness (about 50 nm) was deposited on the spiral-shaped POF by magnetron sputtering to stimulate the SPR effect, thereby forming a spiral-shaped POF-SPR sensor. By modifying the structure of the spiral-shaped POF-SPR sensor, the effects of different structural parameters on the refractive index sensing properties were studied. The experimental results show that the spiral-shaped POF-SPR sensor deforming from a flat POF with a thickness of 500 μm and a thread number of 4 has the better linearity and refractive index sensing characteristics, and the measured sensitivity in the refractive index range of 1.335~1.400 is 1 262 nm/RIU. The proposed sensor has the advantages of low cost, simple preparation and stable structure.

Network terminal (NT) is the data transmission and processing terminal in optical fiber databus, and is also an important part of that. A kind of NT, which intergrated multiple interfaces such as universal asynchronous receiver and transmitter (UART), controller area network (CAN), ethernet and peripheral component interconnect express (PCIE) was introduced. The hardware foundation of this kind of NT was field programmable gate array (FPGA). Through the design of the corresponding intellectual property modules (IP core) of each interface and the design of data-format conversion logic module of different interfaces based on the FPGA platform, the NT design that met the requirements of optical fiber databus was completed. The NT designed in this method could deal with data in different interfaces such as UART, CAN, ethernet and PCIE on optical fiber databus in low time delay. Finally, through the comparison with similar databus products and standards at home and abroad, the results show that the optical fiber databus composed of this kind of NT can effectively solve the problem of data transmission between different types of interfaces.