A denoising technique is proposed based on the temporal-spatial joint analysis to improve the detection accuracy of the LiDAR bathymetry system. Utilize the strong temporal-spatial correlation between echo signals in multiple pulse trigger sampling periods of the LiDAR system， multiple set of similar echo sequences can be found through matching analysis for echo signal of any period. Convert each set of highly correlated echo sequences into a matrix for multidimensional temporal-spatial correlation analysis， and the hard-thresholding shrinkage operator is used in 2-D transform domain to attenuate the noise. Then the basic denoising results are obtained by weighted aggregation. The matching analysis is performed again on the basic denoising data set， and the multidimensional analysis based on the Wiener shrinkage operator is further applied to remove the residual noise. The real signals can be separated from uncorrelated noise. Experimental results demonstrate that the proposed method achieves good effect in noise suppression and edge preservation compared with the widely used LiDAR denoising methods， and improves the peak signal to noise ratio of seafloor signal.

The g-C3N4 nanosheets are prepared by thermal polymerization method using melamine as the raw material. Then CdS nanoparticles are grew by the chemical bath deposition， which in turn construct the g-C3N4/CdS heterostructures successfully. The scanning electron microscope， X-ray diffractometer and X-ray photoelectron spectroscopy are applied to characterize the morphology， crystal structure and chemical composition of samples. It is shown that the CdS nanoparticles with hexagonal wurtzite structure are uniformly attached to the surface of g-C3N4 nanosheets. The g-C3N4/CdS photodetector has a better photoresponse to ultraviolet light at zero bias voltage， and its photocurrent value is about 8 times higher than that of the g-C3N4 nanosheet photodetector. Moreover， the g-C3N4/CdS photodetector also displays well light response in the visible light region， and shows stability and cyclicity of detection to blue and green light.

The persistent photoconductivity of organic material PDVT-10 combined with the polarization field provided by ferroelectric material P（VDF-TrFE）， was used to control the relaxation characteristics of the photosynaptic device by adjusting the polarization intensity of ferroelectric material. The basic functions such as short-term plasticity and paired pulse facilitation of synapses are simulated， and the multilevel dimmable optical synapses are further realized. Moreover， the relaxation phenomenon of continuous photoconductance effect is similar to the flow characteristics of Ca2+ in biological synapses， which can better simulate the synaptic behavior of biological synapses. This study provides a new idea for the development of adjustable photosynapses.

In this paper， the resolution of the super second generation image intensifier and the third generation image intensifier under different illuminations are studied. The results show that the resolution of the super second generation image intensifier and the third generation image intensifier with the same gain， limit resolution， signal to noise ratio and modulation transfer function is constant when the input illumination is above 4.3×10-3 lx， and does not change with the change of input illumination The resolution of the super second generation image intensifier is lower than that of the third generation image intensifier. The lower the input illumination， the greater the difference. The reason is that the third generation image intensifier has higher cathode sensitivity. The higher the sensitivity of the cathode， the higher the initial contrast of the photocathode， and the higher the contrast of the target image on the fluorescent screen， so the higher the resolution under low input illumination. Figure of Merit is an index to characterize the comprehensive performance of image intensifiers. However， when Figure of Merit is used to compare the comprehensive performance of image intensifiers， it can only be compared between the image intensifiers with the same photocathode， but not between the image intensifiers with different photocathode. Therefore， the Figure of Merit can not be used to compare the performance of super second generation and the third generation image intensifiers.

A Two-dimensional （2D） composite grating is proposed to reduce the reflection of the interface of the epoxy-resin-encapsulated silicon solar cell and improve its photoelectric conversion efficiency. The 2D composite grating is prepared by longitudinally stretching One-dimensional （1D） Polydimethylsiloxane （PDMS） grating together with plasma surface modification. Scanning Electron Microscope （SEM） characterization of the 2D composite grating shows that nanowrinkles are longitudinally and orderly distributed on the surface of 1D grating. Optical measurement and theoretical simulation consistently demonstrate the 2D composite grating has superior antireflective property to 1D grating. Electrical characterization shows adding the 2D composite grating onto the epoxy-resin-encapsulated Si solar cell as an antireflective coating can improve the photoelectrical conversion efficiency， reduce the adhesion of dust particles and enhance the surface hydrophobicity and the self-cleaning property of Si solar cell.

When the fiber Bragg grating is multiplexing， the limited source bandwidth may cause spectral overlap， then affects the accuracy of demodulation. A new type of fiber Bragg grating overlapping spectrum demodulation method is proposed. This method is based on an improved manta ray foraging optimization algorithm.The Tent chaotic map is used to optimize the initial population， and the differential evolution algorithm is used to optimize the individual location update strategy， which solves the problem that the manta ray foraging optimization algorithm is easy to fall into the local optimum. The simulation and experimental results of multiple fiber Bragg gratings overlapping spectrum show that the proposed method can accurately demodulate the center wavelength of the overlapping spectrum， even in the case of spectral distortion， the maximum error does not exceed 0.01 nm； at the same time， it effectively reduces the probability of fall into local optimum，and improves the stability and reliability of the algorithm.

The fiber grating strain sensor is developed by using the 3D-printing method. A separate model of strain coupling and transmission between the bare fiber Bragg grating sensor， 3D-printing encapsulation layer and the measured matrix structure is established. The strain transfer relationship between the fiber grating sensor and the measured matrix structure is deduced. The influencing factors of the strain transfer rate of the clamped 3D-printing fiber grating strain sensor are analyzed， including the elastic modulus of the encapsulation layer， the thickness of the encapsulation layer and the bonding length. The research results show that the average strain transfer rate is positively correlated with the elastic modulus and bonding length of the encapsulation layer， negatively correlated with the thickness of the encapsulation layer. The research results can be used as a reference for the clamped fiber Bragg grating strain measurement， error correction and sensor design， as well as the feasibility of 3D-printing technology for packaging fiber Bragg gratings.

Based on previous pattern recognition research which uses Mel Cepstrum coefficient method to extract frequency characteristics of disturbance signal， a fiber intrusion pattern recognition method using 1-Dimension convolutional neural network is purposed for interferometric distributed optical fiber sensing system. Hierarchical thresholds of the restored signal are used to judge and extract the intrusion behavior， which effectively reduces the calculation time compared with the framing method. A one-dimensional convolutional neural network is constructed based on the frequency domain features of the intrusion signal after Fourier transform to extract the characteristics of the disturbance signal adaptively. A line-based Sagnac interference system is set up to acquire data. By training the network with a large number of sample data， a good classification result is obtained. The average recognition rate of the verification set reaches 96.5%. The trained convolution kernels and the convoluted intrusion signals are discussed. After zscore standardization， the one-dimensional convolutional neural network can identify some features in the frequency domain of the signals， and the recognition result of the branch tapping signal with complex frequency components is greatly improved.

The clearance measurement environment of engine and gas turbine is harsh due to the high temperature， high pressure， and strong vibration. In such a measurement environment， the low-coherence heterodyne interferometric clearance measurement technique faces the problem of weak signal and the limited measurement range caused by the low Signal-to-Noise Ratio （SNR）. To address this problem， a method using differential detection is proposed to enhance the SNR. The theoretical measurement model of the proposed method is strictly established， and the model indicates the method has the advantages of improving the SNR and expanding the measurement range. To prove the feasibility of the method， an all-fiber clearance measurement experimental verification system is built， and a comparison experiment between single-ended detection and differential detection is carried out. The experiment result shows that， under the same measurement conditions， the differential detection method improves the SNR by 4.22 times and increases the measurement range from 10 mm to 20 mm. Furthermore， the measurement uncertainty of the system is analyzed. It is theoretically and experimentally found that， due to the scanning speed instability of the optical fiber delay device， the measurement uncertainty increases with the increase of the clearance. But within 20 mm， the measurement uncertainty is less than 15 μm.

Distributed Brillouin optical fiber sensor has great potential for slope monitoring. The sensing fiber is easy to be broken due to extensive local strain， and it is often difficult to obtain accurate measuring results due to the limitation of the spatial resolution. In order to fit for the slope monitoring， an Optical Fiber Sensing Structure（OFSS） for the distributed Brillouin optical fiber sensing network is developed. The OFSS composes of a PVC pipe and two pulleys， constructing multiple round-trips for the sensing fiber. The length of sensing fiber for a short spatial length is extended significantly. So both the sensing accuracy and the tolerance to local deformation of the fiber are improved effectively. A comparative calibration experiment is designed in the laboratory to verify its performance with the spatial resolution of 1 m. What's more， the effect of the proposed OFSS for landslide monitoring is confirmed through an on-site monitoring test on a shallow artificial slope. In addition， since the optical cable has no contact with the outside under the protection of the modified device， the survival rate of the cable is increased significantly.

A high-precision Fiber Bragg Grating （FBG） displacement sensor was proposed based on the combination of a spring and a slider. It can realize the displacement measurement and temperature compensation on a single fiber， which greatly reduces space occupation. Experimental results show that the sensor has excellent micro-displacement measurement capability， its sensitivity， accuracy and the range are 145.08 pm/mm， 1.43%， 1.55% and 10 mm， respectively. The relative error of static synthesis is 2.88%， and the overall error of linearity， repeatability and hysteresis are small. By comparing the temperature compensation effect of aluminum alloy substrate， non-substrate and quartz substrate， it is found that the sensor temperature compensation effect of quartz substrate is better which the delay time is reduced from 6.8 min to 4.3 min， and the maximum temperature compensation error is reduced from 44 pm to 40 pm. Finally， the temperature sensitivity of the sensor made of quartz glass substrate is 6.34 PM/℃ and the temperature compensation error is 0.26%. All of those implies the sensor is suitable for online monitoring of high-precision structural displacements such as mechanical equipment and civil engineering.

A temperature sensing system based on the Fabry-Pérot interference structure is designed. Air， distilled water， 5% NaCl solution， absolute ethanol， methanol and cured silicon rubber are used as the temperature sensitive materials of the sensor to improve the temperature sensitivity effectively. The experimental results show that when the medium in the cavity is air， the temperature sensitivity of the F-P interference structure is inversely proportional to the cavity length. However， when the medium in the cavity is liquid or the solid material constitutes F-P type detection probe， the length of the cavity will hardly affects the temperature sensitivity of the structure. Because at this time the main reason of the wavelength shift is the change of material’s thermo-optical coefficient， the temperature sensitivity is proportional to it. In the experiment， methanol is the liquid with the highest absolute value of the thermo-optical coefficient. And when it is filled with F-P cavity， the temperature sensitivity is -564 pm/℃. In addition， when the cured silicone rubber is used as an F-P type detection probe directly， the temperature sensitivity can be as high as 1.15 nm/℃. The temperature sensing structure has the advantages of small size， good repeatability， strong flexibility and plasticity， and has potential application values in the field of temperature sensing in the future.

In the short-distance， multi-user quantum key distribution integrated optical fiber network scene， aiming at the problem that when the quantum channel and the classical channel are close to each other， the four-wave mixing noise generated by the classical light is easy to have a serious impact on the performance of quantum key distribution， a theoretical model is constructed to describe the deterioration of the performance of discrete variable quantum key distribution caused by the four-wave mixing effect. In order to suppress the influence of four wave mixing noise， four noise suppression schemes were built and analyzed in OptiSystem and MATLAB， which realized the simultaneous co-channel transmission of seven 10 Gbps classical signals and one quantum signal. The minimum bit error rate of classical signal is 8.13×10-11， and the quantum bit error rate is up to 1.69% at the minimum. The research results are of great significance for promoting the practical process of quantum-classical co-channel transmission system.

The traditional convolutional neural network model has substantial spatial feature information redundancy exists in the spatial dimension of the feature maps in hyperspectral image classification， and the spectral band data on a single pixel of the hyperspectral image are regarded as a disordered high dimensional vector for data processing， which does not conform to the characteristics of spectral data， which greatly affects the operational efficiency of the model and the performance of classification. In order to address this problem， a hyperspectral images classification method combined three-dimensional Octave convolution with bi-directional recurrent neural network attention network is proposed. Firstly， the 3D Octave convolution is exploited to capture spatial features of the hyperspectral image，and reduce spatial feature redundant information. Secondly， Bi-RNN spectral attention network is applied to regard spectral bands data as an ordred sequence to obtain spectral information of the hyperspectral image. Then， the spatial and spectral feature maps are connected by means of the fully connected layer to achieve features merge. Finally， the results of classification are outputed through softmax. Experimental results demonstrate that the classification accuracy of the method proposed reaches 99.97% and 99.79% in Pavia University and Botswana datasets. Compared with other mainstream methods， the proposed method can fully exploit spectral and spatial feature information， and own more competitive classification performance.

In order to further improve the speed and accuracy of hyperspectral abnormal target detection， a fast anomaly target detection method based on extended multi-attribute profiles and improved Reed-Xiaoli is proposed. Extended multi-attribute Profiles are extracted from the original hyperspectral images by mathematical morphological transformations. Moreover， a novel fast local Reed-Xiao algorithm is also proposed. Iteratively update inverse matrix of covariance using matrix inverse lemma， thereby reducing the computational complexity of the Mahalanobis distance. The combination of extended multi-attribute profiles and fast local Reed-Xiaoli detector effectively utilizes the spectral information and spatial information of hyperspectral images， it greatly improves the detection accuracy and reduce the running time. Experimental results on three real data sets show the AUC value of the algorithm in this paper is 0.996 7， 0.985 6 and 0.981 6 respectively. The operation time is 21.218 1 s， 15.192 8 s and 32.337 9 s respectively. The proposed method has obvious advantages in detection accuracy and speed， and has good practical value.

To solve the problems of sand-dust image such as fuzzy detail， low contrast and color cast， an enhancement algorithm based on multi-exposure image fusion is proposed. Firstly， the blue channel is compensated to make up for the blue light loss of sand-dust image. Secondly， the RGB three channels of the image are standardized to reduce the deviation between the channel histograms， so as to remove the color cast. In order to obtain the details of different regions in the image， the linear transformation method of parameter control is used to generate multi-exposure images. Weight maps are calculated using quality measures of contrast, saturation and well-exposedness to select the best pixels in the exposure images. Then the Gaussian pyramid of weight maps and the Laplacian pyramid of exposure images are constructed. Finally， the image pyramid is fused and the resulting image is reconstructed. Subjective and objective experiments show that， compared with other algorithms， the proposed algorithm can effectively remove color cast and improve the contrast and clarity of the image， and the result image visual effect is good.

The thermal target information of infrared image and some detail information of visible images are usually ignored in image fusion method based on generative adversarial network. To address the problem， an infrared and visible image fusion method based on dual-path cascade adversarial mechanism is proposed. In the stage of the generator model， a dual-path cascade is used to extract features of infrared and visible images， respectively. To improve the quality of fusion， structural similarity is introduced into the loss function. In the stage of the discriminator model， a dual discriminator is used to distinguish the generated image from the true natural visual images. The proposed method is experimented on the public data， and compared with eight state-of-the-art image fusion methods. The experimental results show that the fusion image not only retains more the target information of infrared images， but also retains more detail information of visible images， which is superior to state-of-the-art methods in subjective evaluation and objective assessment.

The traditional linear variable filter hyperspectral imagers put the linear variable filter in front of the detector window. Due to the difference in the position of the linear variable filter and the detector focal plane， the spectral resolution will be reduced. In order to solve this problem， the hyperspectral imager based on linear variable filter is designed with telescope system， linear variable filter， relay system and detection system. The linear variable filter is placed on the focal plane of the telescope system， and the relay system is to image the linear variable filter on the target surface of the detector so that the linear variable filter coincides with the focal plane of the detector. For the hyperspectral imager system， the spectral range is 400~700 nm， the number of spectrum bands is 31， the spectral resolution is 10 nm， the field of view angle is ±8° and the focal length is 55 mm. After the design and construction of the sample machine， spectral calibration and application experiments is carried out with good results. Compared with traditional hyperspectral imager systems using prisms or gratings to split light， the proposed system doesn't need a collimation system， and has the advantages of small size， light weight and high luminous flux， which can provide a reference for the miniaturization of hyperspectral imagers.

Aiming at dynamic false contour phenomenon may appear when digital driven OLED displays moving pictures， which affects the viewing quality of human eyes， the machanism of dynamic false contour generation is analyzed， and the visual characteristics of the human eye are considered. Just noticeable distortion integral method is proposed to quantify the dynamic false contour generated by different scanning strategies， the reliability of the evaluation method is verified through experiments. The linear pulse width modulation method and the fractal scanning method are combined， the partial fractal scanning strategy is proposed to improve the dynamic false contour phenomenon. When the data bits of the linear pulse width modulation method and the fractal scanning method are both 4 bit， the system verification of field programmable gate array is performed on a silicon-based organic light-emitting diodes microdisplay with a resolution of 1 280×1 024， and the dynamic contour phenmenon may not be perceived by the human eye. Just noticeable distortion integral method is used to evaluate the partial fractal scanning algorithm， compared with the traditional 19 subfields and fractal scanning， the average value of dynamic false contour quantization between any two gray levels is reduced by about 87.86% and 86.51%.

An imaging method that uses multi-layer flat panel detector to acquire X-ray multi-energy images is proposed. Single layer， multi-layer flat panel detector structure is introduced， as well as multi-layer flat panel detector detection imaging system mechanism. X-ray imaging principle is explained， dual energy imaging， subtraction principles are studied by performing single energy X-ray switching simulation with twice shots. Furthermore， different kVp， filters induced X-ray energy spectrum is analyzed， and low， medium， high energy chest phantom images are acquired by performing kVp switching experiment with several shots. Results show that multi-energy images of chest phantom exhibit difference between rib and lung region， and bone enhancement， suppression can be realized by performing dual energy subtract to multi-energy images. Similarly dual energy imaging， subtraction principles are studied by performing dual energy X-ray excited dual layer flat panel detector imaging simulation with single shot. Same kVp with different filter combination induced X-ray energy spectrum is analyzed， multi-layer flat panel detector imaging experiment with single shot is performed， and low， high energy chest phantom images are acquired by first， third layer flat panel detector. Results show that low， high energy images also exhibit difference between rib and lung region， bone enhancement， suppression can also be realized by performing dual energy subtraction. Experiment result proves that X-ray multi-energy imaging using multi-layer flat panel detector detection method is practicable.

Based on the visual advantages of the biomimetic compound eye， the biomimetic curved compound-eye is applied to the unmanned aerial vehicle-borne photoelectric detection system to realize the airborne wide field-of-view and high-resolution detection of moving objects purposes. According to the characteristics of the biological compound eye， a lens array arranged on the curved surface in a hexagon is designed as compound-eye lens， combined with an optical relay subsystem and a CMOS image sensor to form a biomimetic curved compound-eye imaging and velocity measurement system. The developed biomimetic curved compound-eye imaging system has a field of view of 98°×98°， a system focal length of 5 mm， an angular resolution of 1.8 mrad and an F-number of 3.5. The size of the system is Ф123 mm ×195 mm， and the weight is 1.35 kg. According to the imaging principle of the biomimetic compound-eye and by taking advantage of the overlapping field of views between adjacent ommatidia， the velocity measurement principle of the biomimetic curved compound-eye is proposed. The velocity measurement experiment of a moving car shows that the velocity measurement method can effectively improve the test reliability and accuracy of the moving target.

According to the characteristics of the bionic curved compound eye with many sub-eyes and multi-channel imaging， the method of multi-camera bionic curved compound eye calibration and target positioning is carried out. Combining the method of triangulation and camera coordinate system mapping under non-parallel binocular perspective， a method of bionic curved surface compound eye target positioning method is proposed. According to the requirement of compound eye positioning to have a large overlapping field of view between sub-eyes， a 17-eye bionic curved compound eye with a total field of view over 180° is designed with Unity3D simulation software， and a physical prototype is prepared. Through the ZHANG Zhengyou calibration method， the single target calibration of 17 sub-eyes in the compound eye system and the calibration between 38 pairs of adjacent sub-eyes are completed. On this basis， based on the large overlapping field of view between adjacent sub-eyes of the compound eye， the external parameter calibration method of non-adjacent sub-eyes is proposed to realize the unification of the compound eye coordinate system. Through the establishment of an experimental platform， combined with the ORB feature matching method optimized by RANSAC， a compound-eye prototype is used to perform three-dimensional positioning experiments on the unmanned aerial vehicle models at different positions， and then the error analysis of the positioning results is carried out. The experiments prove that the proposed compound eye calibration and target positioning method is applied to the prototype， which can achieve high precision and large field of view target positioning.

Based on the principle of double grating interference scanning， the miniaturized development of the interference scanning type grating encoder is realized by multiplexing transmission grating. A prototype grating encoder was developed suitable for precision displacement measurement. The circuit system of the sensor includes a signal conversion amplifier circuit and a signal processing circuit placed in the D-SUB shell. The 15-pin interface outputs four analog differential signals with a level of 2.5±0.5 V， which can be used for high subdivision to achieve high resolution， With RS422 interface adaptability， it can also improve the anti-interference ability of the sensor in long-distance transmission. The capacitance sensor is used as the displacement reference. The principle prototype of the grating encoder achieves a repeatability accuracy of ±30 nm， and meets the measurement speed requirement of 10 mm/s. The short-term displacement stability of 10 s reaches 5 nm， and the long-term stability of 6 hours reaches 70 nm.

The traditional calibration method for calculating projector distortion is complex in system structure and theoretical derivation. Aiming at this problem， a method of measuring and correcting the distortion of the projector based on a phase target was proposed in this paper. In this method a liquid crystal display was attached with holographic projection film as a phase target. The liquid crystal display sequentially displayed the horizontal and vertical sinusoidal fringe images to obtain the unwrapping phases. The projector projected the horizontal and vertical sinusoidal fringes images to the phase target， then their unwrapping phases were calculated respectively. Using the corresponding relationship between the two sets of phase on each camera pixel， the projection phase of the projector was converted to the liquid crystal display phase coordinate system， then the distortion of the projector can be measured. The distortion was corrected according to the acquired phase space relationship， so that the isophase lines projected by the projector were linearly distributed on the phase target. Experimental results prove that the method proposed in this paper can measure and correct the distortion of the projector without the influence of camera imaging quality. It can improve the projection quality for the fringe projection three-dimensional profile measurement technology.

The morphology of microstructure in glass micromachining is measured with high precision by digital holographic microscopy. The digital hologram of microstructure is recorded based on the Mach-Zehnder off-axis micro-interference system. The high-precision phase distribution is extracted by hybrid reconstruction algorithm and row-column scanning unwrapping. Then the morphology of microstructure is constructed and the depth is calculated. The morphology of flat， convex spherical microlen is simulated and measured， which verifies feasibility of the method. The surface of the glass slide is micro-machined with a carbon dioxide laser， the effects of different lithography parameters on the morphology and depth of glass microstructure is quantitatively analyzed. And the influence of the recondensed height caused by the processing heat affected zone on the processing quality is analyzed. The results show that the lithography depth increases with the increase of laser power and increases with the decrease of marking speed， and the depth is larger when the Q frequency is smaller. This helps guide the fine processing of the glass.

Aiming at the current error problems in the calibration of the optical axis of the high-precision optical system， a method based on the equivalent nodal point theory to calibrate the actual optical axis of the system is proposed. By establishing reference coordinate system， the nodal point coordinate systems and the detector coordinate systems， and combining with homgenours coordinate transformation method， a mathematical model suitable for the optical axis calibration of an actual system is established. The optical system with a 100 mm focal length and a 20 mm distance of optical nodal points is taken as an example to analyze the factors that affect the calibration accuracy， and the result show that the calibration calculation error introduced by collimators， rotary table， and calibration model is less than 10''. It provides a method and reference for the accuracy analysis of optical axis calibration for different optical systems based on the equivalent nodal point theory.

In order to solve the diffraction problem in the spectrum simulation of the star simulator， a method to eliminate the diffraction effect of the Digital Mirror Device （DMD） is proposed. A scalar diffraction model is established， and the DMD is equivalent to a two-dimensional blazed grating. Based on the grating diffraction theory， the factors affecting the diffraction energy level are searched for， and the coupling relationship between the incident angle and the diffraction effect is deduced. When the light beam is incident along the diagonal of the DMD and the height angle is equal to the angle of height， the diffraction efficiency is maximized， and the diffraction efficiency is increased by nearly 12%. At this time， the diffraction loss energy is the least， and the diffraction effect when the DMD reflects is minimized. Through ZEMAX Optimal design of Czerny-Turner optical structure to correct system aberrations. The optimization results show that the spectral simulation accuracy is better than 5% in the 500~900 nm spectral range， and the spectral resolution of the full spectral range is better than 5 nm， which can effectively improve the ground calibration accuracy of the star sensor.

A correction method is proposed to improve the star map simulation accuracy of splicing star simulator， which overcomes the limitation of conventional methods in improving star map accuracy merely through correcting the principal distance. The aberration of a small distortion flat field high imaging quality collimating optical system is optimized after analyzing the working principle and usage of splicing star simulator. Based on aberration design results and the star position measurement model， the correction models for distortion， coma， curvature of field are created separately. In addition， a device for measuring star position accuracy is built by utilizing the theodolite and six-dimensional adjustment mechanism. According to measurement results， the correction model based on aberration theorise effective in improving the star map accuracy of splicing star simulator. The maximum star position errorr educes from 48.78" down to 10.75"，meetting the technical requirements.

Based on optical phase-locked loop technology， a Raman laser system including three lasers，reference， master and slave lasers， is studied. Among which， the reference laser was firstly frequency-locked by modulated transfer spectrum technology and then as a standard frequency； the master laser， which has a frequency difference of 1 GHz relative to the reference laser， was red detuned from excited stated to prevent direct interaction between lasers and atoms； the salve laser， having a frequency difference of 6.8 GHz relative to the master laser， was used to excite the transmission between the hyperfine energy levels of ground state 87Rb atoms. Respectively， the master and slave lasers were locked by two sets of optical phase-locked loop. The measurement results reveal that the phase noises of the two sets of optical phase-locked loop are within the range of 100 Hz to 1 MHz and lower than -70 dBc/Hz and -65 dBc/Hz respectively， and accordingly， the influence of phase noise on atom interferometric gravity measurement is as low as 5.3×10-7 per shot. The master and slave lasers are amplified by two separate taped-amplifiers to ensure that the master laser’s power is half to that of the slave laser； the composition， polarization unification， and frequency modulation of both the master and slave lasers are realized by the combination laser transmission in fiber and free space. As a result， the Raman laser with a total power of 180 mW is realized， and the maximum fluctuation of total power less than 5% is achieved in long-term power measurement depended not on any additional power stabilizer， satisfying the experimental requirement of atom interferometry measurement.

Based on using the Acousto-optic Modulator （AOM） as a feedback device， we design the power locked setup of the 780 nm laser in the common band of atomic clocks. Through low-noise design and parameter optimization， the suppression of key sources of the noise and the laser power locking are achieved. In the frequency deviation range of 20~10 000 Hz， the relative intensity noise of laser beam is effectively suppressed. Especially at the frequency of 1 300 Hz， 20 dB suppression of RIN has been reached. At the same time， the relative power stability in the mid-term （9 000 s） has been increased 78 times from ±0.754% to ±0.009 68%. According to the typical parameters of CPT atomic clock， the influence of relative intensity noise of the laser on the frequency stability of an atomic clock is only 2.1×10–14@1 s. In addition， the mid-term stability improved from 4.67×10-2 in unlocked state to 8.67×10-5 in locked state，and it will be helpful to improve the intermediate period frequency stability of the atomic clock.

The analytical expression for the cross spectral density function of partially coherent circular edge dislocation beams propagating in the deep dermis of mouse tissue is derived based on the generalized Huygens-Fresnel principle， the effects of the initial beam parameters （the beam wavelength λ and the number of circular edge dislocations n） and the propagation distance z on the normalized intensity distribution， phase evolution and propagation trajectory of the beam are investigated. The results show that the central intensity of the partially coherent circular edge dislocation beam with n dislocation number is the largest in the source plane， and 2n secondary peaks symmetrically distribute on both sides. With the increment of propagation distance， the intensity distribution gradually evolves from multi-peak to single-peak， the longer the wavelength is， the smaller the n is， the faster the intensity distribution evolution is. The more the number of dislocations is， the better the beam stability is. Additionally， in the source plane the radius of the innermost ring of n circular edge dislocations decreases as the number of dislocations increases. Owing to the combination of the biological tissue turbulence induction and diffraction effect， the n circular edge dislocation has split into n sets of coherent vortices whose topological charges are "+1" and "-1" from the beginning of the propagation， respectively. As the propagation proceeds， n sets of coherent vortices with topological charges of "+1" and "-1" will be generated. The bigger the wavelength is， the more the n is， the faster the evolution of the beam phase， the more the distribution of the coherent vortices tends to be concentrated， and finally all the coherent vortices are annihilated. The smaller the distance between the coherent vortices， the earlier the annihilation. The longer the wavelength is， the larger the value of n is， the earlier the annihilation of the initial coherent vortices pairs is， and the longer the propagation distance to the total annihilation is. The smaller the wavelength is， the larger the value of n is， the earlier the annihilation of the newly generated coherent vortices pairs is， and the longer the propagation distance to the total annihilation is.

A new algorithm for iterative optimization of wavefront shaping which is the combination of Nondominated Sorting Genetic Algorithm Ⅱ-Hybrid and the superpixel method is proposed， using digital micromirror devices for light field modulation. Multi-point controllable light focusing of reflective wavefront shaping is realized. Our method has two advantages. On the one hand， while increasing the enhancement factor， it also ensures that the uniformity of the intensity of multiple focal points is controllable， that is， all focal points have uniform intensity； on the other hand， the structure of reflective wavefront shaping is more conducive to application. In order to verify our algorithm， we use frosted glass as the scattering media in the experiment. The experimental results show that， compared with the genetic algorithm based on superpixel method， the enhancement factor of the focal point is increased by 24.12%， and the coefficient of variation is reduced from 13.3% to 4.2%. This research provides a new method for reflective wavefront shaping multi-point light focusing， and has potential application value in the field of optogenetics and light capture.

The electron temperature of the Ar/CH4 plasma was diagnosed under the pressure of 40~80 Pa and the microwave power of 400~800 W by Optical Emission Spectrometry （OES）. The experimental results show that the electron temperature obtained by the OES is between 0.75 eV and 4 eV under the above experimental conditions. Through measurement of Ar or CH4 plasma， it is feasible to use the OES method in microwave coaxial plasma with carbon-containing gas. This research can further expand the application of the OES method in the PECVD field.

In order to detect the concentration of methane in water， a sensor system of dissolved methane in water based on off-axis integrated cavity output spectroscopy is developed in this paper. A distributed feedback laser （center wavelength， 1 653 nm）， a laser temperature control module， a laser current drive module， a resonator cavity/gas chamber， a photoelectric detector， a data acquisition module， a data processing module and a gas-liquid separation module were included in the system. Experiments including calibration of effective optical path length， calibration of direct absorption signal， and stability test were carried out by using methane gas samples and pure nitrogen （N2）. The effective optical path length of the cavity is calibrated using a methane sample with a concentration level of 10×10-6， which is determined to be 1 906 m. Using pure nitrogen as the target gas， the stability of the system is measured. The results of Allan variance analysis show that when the averaging time is 2 s， the detection sensitivity is 92.8×10-9， and the sensitivity of the system can be improved to 13.2×10-9 with an averaging time of 134 s. The system was used to measure the concentration of the dissolved methane in tap water， rainwater and lake water， and the results confirmed the engineering practical value of the technique and system. The research work and related results have laid a good foundation for water quality detection and exploration of clean energy such as natural gas hydrate.

In order to expand the detection range， the 2 μm band laser generated by thulium-doped fiber laser is used to conduct intracavity gas sensing of water vapor in human breathing gas. Firstly， the direct absorption gas sensing technology is analyzed theoretically. Secondly， the characteristics of thulium-doped fiber pumped by a 1 570 nm laser are studied， and the spontaneous emission spectrum is mainly distributed in the band of 1.85~2.05 μm. Then， an all-fiber thulium-doped fiber ring laser is built， further using a tunable filter， the wavelength output in the range of 1 927.5~1 985 nm is achieved and the laser linewidth of 0.05 nm is obtained， which shows the advantages of single longitudinal mode， narrow linewidth and stable output. Finally， combined with wavelength sweep technique， the spectral scanning of water vapor in the range of 1928~1938 nm is realized and eight absorption lines are resolved， which are consistent with the simulated spectrum based on the HITRAN spectrum database. The absolute error of wavelength positioning is less than 0.03 nm. The results show that the intracavity gas sensing system is suitable for 2 μm band gas sensing detection.

Terahertz time domain spectroscopy technique was used to detect the defects of high temperature resistant composite materials with multi-bonded structures. In order to identify the debonding defects in both upper and lower layers at the same location， the terahertz signal waveforms in the non-defect area， upper and lower debonding areas were analyzed. The characteristic peak-to-peak， skewness， minimum value， peak-to-valley value， waveform factor and absolute mean value of signal amplitude were taken as the input of BP neural network.The initial weight and threshold value of BP neural network were optimized by Particle Swarm Optimization （PSO）， which solved the problem that BP neural network was easy to fall into local optimum. The optimized PSO-BP neural network could realize the identification of the debonding defects of upper 100 μm and lower 100 μm， with the accuracy of 90.71% and 86.92%.