A new kind of polarization maintaining photonic crystal fiber with two side-holes is designed. There is only a layer of small air holes between the core and the large air hole， and two large air holes are introduced in the cladding symmetrically. The two-dimensional model was simplified by the plane strain hypothesis， the two-dimensional model of this fiber was numerically analyzed by the finite element method， and the characteristics of temperature and hydrostatic pressure sensing were studied by calculating the birefringence-induced frequency shift at different temperatures and hydrostatic pressures. As the studies show， within a large range of hydrostatic pressure and temperature， this polarization maintaining photonic crystal fiber can realize the hydrostatic pressure sensitivity of -2.135 3 GHz/MPa and it is insensitive to temperature without doping any stress material. Its temperature sensitivity is only +0.154 2 MHz/℃. In addition， the optical properties of the photonic crystal fiber were also analyzed. The proposed photonic crystal fiber satisfies the condition of single-mode transmission， has a small confinement loss and a large effective mode area. Due to its characteristics of small size， strong compatibility with other optical fibers， high hydrostatic pressure sensitivity and temperature insensitivity， it has obvious advantages in the accurate measurement of hydrostatic pressure in the environment with variable temperature and great hydrostatic pressure variation. Better optical properties make it has important reference value in monitoring application of oil well and civil engineering.

Fiber Bragg Grating （FBG） thermal flow sensors are only suitable for gas flow currently. In order to expand its application field， a new FBG thermal flow sensor that can be used for liquid flow measurement is designed. A heating ceramic sheet is used to provide heat at a constant power for the FBG thermal flow sensor. Different flow of liquid takes different heat when flow through the sensor. By detecting the change of FBG central wavelength， the temperature change of the sensor can be measured， and then the liquid flow can be deduced. Through temperature sensor test experiment and flow sensor test experiment， it is verified that the designed sensor can be used for liquid flow measurement. The experimental results show that the flow measurement range of the sensor is 40.575~550.664 L/h.

Aiming at the problems of severe noise interference of fiber-optic vibration signals， single feature extraction and long recognition time， an improved local characteristic-scale decomposition and ant colony algorithm optimize deep belief network are proposed. Firstly， cubic B-spline function interpolation is used to fit the mean curve to improve the local characteristic-scale decomposition algorithm， and the sum of a series of intrinsic scale components is obtained by decomposing the original signal. Secondly， the fusion index is formed by kurtosis factor and energy spectrum coefficient to screen the effective component. Then， the entropy features of the effective components in the time domain， frequency domain and time-frequency domain are extracted respectively to perform feature fusion and dimensionality reduction. Finally， the integrated feature vectors are fed into ant colony algorithm optimized deep belief network for training and recognition to improve the algorithm efficiency and recognition rate. Experimental verification using measured data shows that the signal-to-noise ratio is increased by 8 dB on average， the average signal recognition rate can reach 95.83%， and the average recognition time is 0.715 s.

A thin-cladding Excessively Tilted Fiber Gratings （ExTFG） cantilever beam vibration sensor is reported in this paper. Using the ExTFG written in the standard single-mode fiber， the effects of the decrease of the fiber cladding radius on the dispersion factor of the waveguide， the effective refractive index of the cladding mode， the axial strain sensitivity factor， the axial strain sensitivity and the mode order are theoretically analyzed， and the corresponding numerical simulations are performed， which provide the theoretical foundation for the enhancement method of vibration measuring sensitivity. And then， hydrofluoric acid is used to etch the fiber cladding to fabricate ExTFGs with different diameters and the related vibration sensing experiments are conducted. The results show that in the vibration frequency range of 40~200 Hz， with the decrease of cladding diameter， the acceleration sensitivity of ExTFG vibration sensor in C band of the same order and different order TE/TM mode increases gradually， and there is a good linear relationship between them. The maximum acceleration sensitivity of the TE and TM mode with the same-order cladding mode is 100.46 mV/g and 88.68 mV/g， respectively， which is increased by 1.36 times and 1.53 times， respectively， as compared with the standard diameter one. And the maximum acceleration sensitivity of those with different-order cladding modes can reach 159.35 mV/g and 133.37 mV/g， respectively， which is increased by 2.15 times and 2.31 times.

Convolutional neural network-based semantic segmentation models do not effectively explore feature weight information， which will result in under-segmentation of segmentation boundaries in complex scenes of computed tomography images. To address this problem， an improved U-Net++ model is proposed to explore adaptive weighted aggregation strategy based on U-Net++， and the improved U-Net++ model is applied to the segmentation of lung nodules in computed tomography images. In the convolutional neural network phase， the information from the different levels of deep features is extracted and combined with the weighted aggregation module， and thus the weights of features in each layer are adaptively learned. Then the learned weights are loaded on each feature layer and obtained a sampled segmentation map， and the final segmentation result can be obtained. Segmentation experiments are carried out on the lung cancer data sets of LIDC and Chongqing University Cancer Hospital. The intersection over union of the proposed improved U-Net++ method on two datasets reach 80.59% and 87.40%， and the DICE of this method on two datasets could reach 88.23% and 90.83%， respectively. Compared with U-Net and U-Net++， the proposed algorithm significantly improves the segmentation performance of lung nodules in computed tomography images. The experimental results show that improved U-Net++ achieves accurate segmentation on tiny details of tumors， and it bring beneifits to solve the problem of under segmentation when lung nodules grow invasively to the surrounding.

A dual channel residual convolution neural network with independent weight is established. The features of target in visible and infrared images is extracted.The multi-scale composite frequency band feature maps are generated. Based on the Euclidean distance between image points， the saliency of each image point in the dual band feature map is calculated. The adaptive fusion is carried out according to the characteristic contribution value of the target in different imaging frequency bands. Through the thermal radiation pooling kernel and visual attention mechanism， the logical mask of the target region of interest under dual frequency band is generated and superimposed on the fusion image to highlight the target features and suppress the non target area. Based on end-to-end identification network and using the cross loss calculation strategy. The target recognition of multi-scale dual band fusion feature map with attention mask is carried out.The results show that the designed recognition network can effectively integrate the physical characteristics of infrared heat source and the line features of visible image. The depth of information fusion is improved. The thermal radiation and texture features of the target is retained. The interference of background information is reduced. It has good recognition accuracy and robustness for multi-size heat source targets in all-weather and complex environment.

In view of the fact that the mountainous area is easy to accumulate clouds and haze， it brings difficulties to the interpretation of remote sensing images. In addition， the remote sensing image contains a large amount of information and the calculation rate is slow. In the process of dehazing， aiming at the problem of poor removal effect when the amount of haze is too much. A dehaze algorithm for optical remote sensing images that combines curvature filtering optimization and non-local dehaze method is proposed. First， curvature filter is used to reduce the total energy of the images， thereby reducing the memory occupied by the images and improving the calculation efficiency. Then the Hough transform voting mechanism is used to screen out the corresponding atmospheric light value， and the transmittance is estimated according to the atmospheric scattering model. Finally， the image reconstruction method is used to obtain the dehazed image. The experimental results show that the calculation rate of the proposed method is significantly faster than the method without Gaussian curvature filtering optimization. In terms of dehazing effect， the restored images are more detailed and clearer when the amount of haze is large， and the brightness is appropriate， which improves the effect of dehazing.

Because the existing unsupervised band selection methods in hyperspectral image classification could not calculate the similarity between bands and the high-dimensional characteristics in the selection process， a greedy unsupervised hyperspectral band selection method based on variable precision rough set was proposed. Firstly， a new dependency measure was defined by using the variable precision rough set， which made it insensitive to the misclassification parameters of the variable precision rough set， so as to make full use of the similarity between wavebands. Secondly， a new criterion was proposed to find out the bands with higher and lower similarity values in the unselected and selected segment subsets. Then， the first order incremental search method was used to select the required information band one by one， so as to avoid the generation of a large amount of information and reduce the computational complexity. Finally， three hyperspectral datasets were used to compare the proposed band selection technique with the five latest techniques. The results show that the proposed method has good classification accuracy for all datasets， and the average classification accuracy is only 1.9%，3.1% and 4.1% lower than the average classification accuracy of all pixels under the condition of 50% marked pixels， which proves that the proposed method can guarantee good classification performance and generalization ability of data set， and has robustness to parameters.

The status and difficulties of infrared interferometry imaging are analyzed， the principle of infrared coherent detection with laser local oscillator is introduced， the principle of infrared spectral segmentation and interferometry imaging based on electronics is expounded， and the frame structure of laser local oscillator infrared array detector is discussed. By means of laser local oscillator and coherent detector， the correct phase transmission of the infrared signals of the two telescopes can be ensured， and the narrow-band filtering in electronics can be implemented. The narrow-band infrared signals are favorable for the long-baseline interferometry imaging. On this basis， similar to the microwave synthetic aperture radio telescope， it can combine many small apertures in different spatial positions to form a large optical aperture， in the form of infrared spectral "radio" telescope to achieve high-resolution astronomical imaging， which is likely to greatly reduce the complexity， volume and weight of the infrared imaging system. The characteristics of the stratosphere airship platform are described， which provides favorable conditions for the installation of long-baseline large diffraction aperture telescopes and can greatly reduce the impact of the atmosphere on astronomical observation， and may become a new platform for astronomical observation. The arrangement scheme of the infrared spectral interferometry imaging telescope with 10 m baseline and 2 m diffraction aperture is presented， its detection and imaging performance is analyzed， and the key technologies and possible technical approaches are discussed. The analysis shows that， the infrared astronomical observation capability of the 10 m aperture telescope can be equivalent through three 2 m aperture diffraction telescopes with the 10 m baseline based on the stratospheric airship platform.

Non-linear partial volume effect is still an unsolved problem in the theory and application of computed tomography reconstruction. In this work， an optimization-based reconstruction algorithm for effectively solving this effect was studied and developed. The discrete non-linear X-ray projection transformation model was established， when the inverse problem of X-ray projection transformation was transformed into a non-convex optimization. A non-linear iterative reconstruction algorithm was formulated to solve the non-convex optimization problem by tailoring the first-order primal-dual algorithm developed originally for solving convex optimization problems. Using computer-simulation studies， the convergence and reconstruction accuracy of the algorithm was demonstrated in this investigation. Images reconstructed were examined first through visual inspection with extremely narrow display window for revealing a contrast level of less than 1%， and then assessed with quantitative metrics such as the normalized l2-norm of the difference relative to its truth images. The simulation results show that the non-linear reconstruction algorithm can converge to the real image within the range of calculation accuracy， and the reconstructed image can also be displayed for the details with image contrast less than 1%. This would provide a reference for the design of computed tomography imaging application algorithm which can effectively compensate the non-linear partial volume effect artifacts.

Aiming at the problems of complicated background gas composition， difficult detection， and poor stability of methane in the atmosphere， based on tunable diode laser absorption spectroscopy technology and wavelength modulation spectroscopy technology， the cascaded integrator comb filter and finite impulse response filter are applied to the digital quadrature lock-in amplifier to investigate experimental research on trace detection of methane gas in the atmosphere in this paper. Experiments show that compared with the traditional digital quadrature lock-in amplifier， the signal-noise ratio of the second harmonic signal extracted by the improved digital quadrature lock-in amplifier is improved from 38.61 dB to 44.95 dB； Applying the nonlinear iteration partial least square-extreme learning machine algorithm model to the methane gas concentration inversion， compared with the classic least square method， the root mean square error decreases by 0.907. Through 16 sets of concentration step-by-step experiment tests， the actual detection limit of the system is 1 ppm； The long-term stability test is conducted at 600 mbar pressure， 25 ℃ and 50 ppm methane concentration for 3 hours. The variation range of methane concentration detected is 49.6 ppm~50.3 ppm， and the standard deviation is 0.092 1 ppm. When the integration time reaches 56 s， the theoretical detection limit of the system is 25.6 ppb. This shows that the cascaded integrator comb filter and the nonlinear iteration partial least square-extreme learning machine algorithm model have higher superiority and practical prospects in infrared gas detection.

In order to solve the problem of manual calibration in the existing laser scanning projection system， the self-calibration method of laser scanning projection system combined with monocular vision was proposed. Firstly， laser scanning projection system model combined with monocular vision technology is established， and the self-calibration method of laser scanning projection system combined with monocular vision is studied. Secondly， the mathematical model between camera and laser scanning projection system and projected object is founded. Finally， it is proved that the self-calibration method of laser scanning projection system combined with monocular vision can solve the problem by the simulation experiment. At the same time， the accuracy of the self-calibration method is verified to meet the requirement of the laser scanning projection system by comparing the experimental results of Matlab and SA. The relationship between accuracy and external factors such as distance and angle is analyzed. The simulation results show that when the distance is fixed and the rotation angle of the camera is 55°， the range of the camera from the projector and projected object is 1.5~3 m and meet the accuracy requirement of the projector. By independently building a new laser scanning projection system， the studied method has practical operability and feasibility.

On the same platform， two or more different polarization loads are used for joint observation， achieving complementary advantages by fusion inversion so as to obtain data products with higher precision and quality. To achieve the joint detection and cross calibration transmission of two polarization loads， the field of view matching between them is one of the key issues that need to be solved. For this purpose， an aviation verification system based on the Particulate Observing Scanning Polarimeter（POSP） and the Simultaneous Imaging Polarization Camera（SIPC） is built， then the flight experiment is completed. According to the sensing geometry and spatial response characteristics of two-polarization instrument， the corresponding models of observation pixel and geographical spatial position are established by coordinate transformation， meanwhile， the geolocation result is corrected by the digital elevation model and the instrument misalignment correction. The error sources that affect the accuracy of geolocation are analyzed. Based on this， an error statistical model based on Monte Carlo method is established for simulation calculation. This method is applied to airborne SIPC and POSP geolocation， and the consistency of the measurement results between the two instruments is analyzed. The results show that the polarization degree deviation of the two polarization instruments is less than 2% and the radiance deviation is less than 5% after the geospatial locations is matched， indicating the feasibility and validity of the geolocation and correction method. It provides an effective way for the subsequent geolocation of spaceborne polarized crossfire detection.

In order to improve the accuracy of star centroid extraction of high dynamic star sensor， an on-orbit correction method based on adaptive filtering was proposed， which can adaptively adjust the correction matrix at different angular velocities. Correction matrix is updated in real time by the noise estimation filter which is proposed based on temporal-spatial correlation. And the star point can be corrected in real time. Compared with the traditional method which correct on the ground， it not only reduces the cost， but also updates the correction matrix in real time， so the correction is more reliable. Experimental results show that the proposed method is effective in correcting star sensor under high dynamic conditions. Compared with the existing on orbit correction methods which can only appicable to low-speed star tracker （≤0.1°/s）， the proposed method can adapt to the angular velocity of 0~3°/s， accurately correct the high dynamic star sensor， and successfully improve the accuracy of star centroid， which is of great significance to improve the dynamic performance of star sensor.

The frequency-selecting characteristics of a triple-ring passive resonator are analyzed by Jones matrix， and an erbium-doped fiber laser based on the triple-ring passive resonator is designed according to the frequency-selecting theory of the passive resonators. The experimental results show that the triple-ring passive resonator has a good frequency-selection characteristic. When the pump power is 159 mW， the stable single-longitudinal-mode lasing signal is obtained. The maximum fluctuation of output power is 0.04 dB and the maximum variation of output wavelength is 0.016 nm. The coupling ratios of the optical couplers decrease， the laser will be in multi-longitudinal mode oscillation.

An approach for recognizing spatter dynamics and analyzing welding status based on spatter feature is proposed. A type 304 austenitic stainless steel plate is taken as a testing object for high-power disk laser bead-on-plate welding experiment. A high-speed camera is used to capture the ultraviolet band and visible light band spatter images. Image processing extracted spatter feature parameters， including centroid position， area， grayscale， average grayscale， and radius. Spatter searching information database and similarity functions are established based on spatter feature parameters to recognize the spatter， calculate spatter volume and grayscale， and evaluate the welding status. By comparing the weld seam width with the spatter feature information， the internal relation between the welding status and the spatter feature parameters is investigated. The test results of the relationship between spatter characteristics and welding state show that the weld width will decrease with the increase of spatter volume and gray value in the process of laser welding. It can be seen that the status of a high-power disk laser welding process could be monitored and evaluated through the spatter feature parameters.

A novel Huygens particle， of which transmission magnitudes and phases can be controlled independently， is proposed here. The transmission magnitude can be arbitrarily tuned from 0 to 0.9， while the transmission phase is between 0 and 2π. Aligning Huygens particles based on generalized snell’s law， a metasurface structure with function of anomalous transmission is designed and fabricated. Full-wave simulation results prove that the plane electromagnetic wave normally incident from the bottom is nearly steered to 60° by the designed Huygens metasurface at 10 GHz. The near-field measurement results are basically consistent with the simulations， which demonstrates the correctness of desinged metasurface further.

The plasma polishing and etching system independently developed by the laboratory was used to prepare sapphire self-organized nanostructures. The formation mechanism of self-organized nanostructures by Ar+ ion beam sputtering assisted with stainless steel at different incident energy was studied. The roughness， longitudinal height of nanostructure and surface morphology of the etched sapphire sample were respectively measured by Taylor Surf CCI2000 non-contact surface measuring instrument and atomic force microscope. With the ion beam incident angle of 65°， the beam current density of 487 μA/cm2， the etching time of 60 min， and the ion beam incident energy of 1 000 eV， well-ordered stripe-like nanostructures with a longitudinal height of 11.1 nm appear on the surface of the sapphire sample. As the incident energy increases， island-like nanostructures gradually appear on the surface of the substrate. When the incident energy is 1 200 eV， island-like structures and ripple generate on the substrate， with a longtidudinal height of 13.6 nm. With an increase of the incident energy， the density of island-like structures enhances. When the incident energy is 1 400 eV， the longitudinal height of island-like nanostructures is 18.8 nm.When the incident energy is 1 600 eV， the surface of the sapphire sample appears a relatively ordered island-like structure with a longitudinal height of 20.1 nm. During the ion beam sputtering process， with an increase of the incident energy， metal impurities break the balance between the surface growth mechanism and the surface smoothing mechanism， which form island-like nanostructures， these structures promote the growth of nanostructures and change the orderliness of nanostructures.

The spectral response， cathode sensitivity， gain， resolution and signal-to-noise ratio of the super second-generation and the third-generation image intensifier were measured at the short wave cut-off of 550 nm， 625 nm and 675 nm for A luminant. Under 10-1 lx illumination， the resolution of the third-generation image intensifier does not decrease after undertaking 675 nm short wave cut-off for A luminant， while the resolution of super second-generation image intensifier drops to 94% of the initial value； however， under 10-4 lx illumination， the resolution of the third-generation image intensifier decreases to 90% of the initial value after undertaking 675 nm short wave cut-off for the A luminant， while that of the super second-generation image intensifier decreases to 85%of the initial value. In addition， the resolution decreases with the decrease of illumination. For the super second-generation and third-generation image intensifiers， although the parameters are the same under the illumination of A luminant， but when used in different short wave cut-off wavelengths， the performance of the third-generation image intensifier is better. Although the decline proportion of parameters for super second-generation image intensifier is higher than that of third-generation image intensifier under different short wave cut-off wavelengths， the difference is not obvious， so the performance difference during use is not obvious.

The polarization remote sensing experience threshold cloud detection algorithm is strongly affected by subjective factors， and it is very easy to have the problem of inaccurate cloud detection over bright ground. In response to this problem， this paper proposes a machine learning cloud detection algorithm that combines active and passive remote sensing satellites. The algorithm is based on the multi-channel multi-angle polarization characteristics of the POLDER3 payload and the high-precision cloud vertical characteristics of the CALIOP payload. It uses POLDER3 payload and CALIOP. The load observation overlaps the regional data， and the BP neural network optimized by the Particle Swarm Optimization algorithm is built to train the cloud detection model. Based on the cloud detection training model， a cloud detection experiment was carried out using POLDER3 level-1 data. The experiment showed that the cloud detection result of this algorithm is 92.46% consistent with the MODIS cloud detection product， which is higher than the consistency between the official POLDER3 cloud detection product and the MODIS cloud detection product 83.13%. By comparing the experimental results of the algorithm in this paper with the optical characteristics of different pixels from the official POLDER3 cloud detection product， it is found that compared with the official POLDER3 algorithm， this algorithm is more sensitive to thin clouds over the bright surface and can perform cloud detection more effectively.

To improve the detection performance of the photoacoustic cell， a trapezoid compound photoacoustic cell is proposed and analyzed. Taking the traditional cylindrical resonant photoacoustic cell as the reference model， the reliability of the simulation method is verified by comparing the results of analysis and simulation. Based on the simulation method， the influence of the radius， length， and the number of the stepped cavity on the acoustic field and flow field characteristics of the trapezoidal composite photoacoustic cell is calculated. The results show that reducing the stepped cavity radius of the trapezoidal compound photoacoustic cell can enhance the photoacoustic signal， and the optimal length of the stepped cavity can make the photoacoustic signal to be the strongest， and the number of the stepped cavity should be one. In terms of the flow field， the configuration characteristics of the trapezoidal composite photoacoustic cell improve the condition of gas vortex return to the cavity. If the transition to the cavity is furthering rounded or chamfered， the velocity gradient in the cavity will become more stable. Select a set of design parameters， the calculation results show that the volume of the trapezoid compound photoacoustic cell cavity is reduced to 39.7% of the corresponding cylindrical resonant photoacoustic cell， and the photoacoustic signal is increased by about 18.7%. In addition， its frequency response bandwidth is narrowed， and its quality factor is improved. The overall results show that the acoustic field and flow field characteristics of the trapezoidal composite photoacoustic cell are better than that of the corresponding cylindrical resonant photoacoustic cell. The research content can provide reference for the structural optimization and improvement of the photoacoustic spectral photoacoustic cell.