By annealing at high temperature (850~950℃), the fiber Bragg grating can be erased at high temperature and grow again to form a regenerated fiber Bragg grating, which can work stably in a high temperature environment of more than 1 000℃. However, the mechanical strength of the regenerated fiber Bragg grating after annealing at high temperature is significantly lower than that of common fiber Bragg gratings. In this paper, the axial stress of fiber Bragg grating and the change of quartz molecular components in fiber Bragg grating are studied and analyzed by single mode quartz fiber experiment. The results show that compared with the unannealed fiber Bragg grating, the compressive stress at the fiber core decreases by 80 MPa, and the tensile stress at the cladding far away from the fiber core gradually decreases by 22 MPa. At the same time, with the increase of oxygen content in the atmosphere of hot annealing, the regenerated fiber grating SiO2 generated after annealing gradually increased, and the proportion increased from 52.99% to 69.92%. Although SiO2 has a high density and its mechanical strength is greater than Si2O3, the brittleness of the regenerated fiber grating after hot annealing still increases. Therefore, it is inferred that the change of components has no significant effect on the increase of brittleness of regenerated fiber Bragg grating, and the main reason for the increase of brittleness is the stress relaxation caused by high temperature. This paper provides a reliable theoretical and experimental basis for improving the mechanical properties and solving the brittleness problem of thermal regenerated fiber Braggg ratings.

The forward stimulated Brillouin scattering based fast light in small-core photonic crystal fibers is theoretically investigated. Three-wave coupled wave equations of forward stimulated Brillouin scattering in frequency domain were derived to calculate the group refractive index and gain coefficient in small-core photonic crystal fibers by Fourier transformation, then optical and acoustic field distribution, advancement and broadening factor of signal pulses induced by forward stimulated Brillouin scattering were simulated by the finite element method. Tight confinement of the optical fundamental mode and acoustic modes strengthens nonlinear interaction in the small-core photonic crystal fibers and results in strong SBS and large advancement of time. The time advancement grows nonlinearly with the transmission distance of signal light increasing, and the signal pulses are compressed. The pulse broadening factor gradually levels off with the growth of the initial pulse width. The time advancement of 21.76 ns and pulse broadening factor of 0.77 are evaluated at the transmission distance of 70 m, the initial pulse width of 200 ns and the pumping pulse power of 600 mW.

A nonlinear reconstruction method with a single digital hologram using deep learning was proposed for off-axis Fresnel digital hologram. Classic Fresnel diffraction integral is utilized for simulating digital holographic imaging to provide the training samples, and a deep convolution residual neural network is utilized to implement on the object image reconstruction from the recorded hologram, by learning the nonlinear mathematical mapping from the digital hologram to the corresponding object image. The results of numerical simulation experiments show that the method could directly eliminate zero-order images and twin images without fringe pre-processing procedure for extracting object term, compared with the traditional frequency filtering and four-step phase-shift techniques for achieving Fresnel digital holography reconstruction, as well as high quality reconstructed object image. It also has strong robustness to the test dateset generated with different diffraction distances using same recording reference light waveform.

Most of the image dehazing algorithms based on physical models have the problems of restoring image color distortion and halo effects in the boundary area of the sky. In this paper, we propose an image dehazing method that based on light field depth calculations with atmospheric scattering models. In this method, the depth of the scene is calculated by the light field epipolar plane image, and the initial transmission calculated by the scene depth information is fused with the dark channel transmission to obtain the final transmission. At the same time, the depth of the scene is used to determine the sky boundary, and the sky area is processed separately. The experimental results on the synthetic haze image and the real haze image show that the proposed method has both improve Peak Signal-to-noise Ratio(PSNR) and Structural Similarity(SSIM) compared with a variety of methods. This method can also get good result in recovery the color of haze image and reduce halo effect after dehazing.

In order to avoid the characters of current double-random phase-encoded optical image encryption, whose key and ciphertext are large in size, and has weak ability to resist the attack of selecting plaintext and ciphertext, a lossless compression optical image encryption method was proposed based on Logistic-Logistic cascade chaos and vector decomposition. Firstly, the plaintext is divided into two images by space sampling and scrambling. Then the images are synthesized together by the interference. At last, the ciphertext can be gotten by putting the image into a double random phase encoding system. The scrambling improves the security of the encryption system, because it can avoid plaintext information to be seen after interference. The interference reduces the volume of the ciphertext to about the half in size, which is convenient for ciphertext transmission. The decryption method based on unit-mode vector decomposition could avoid the problem of low resolution of the decrypted image caused by the existing compression method. Logistic-Logistic cascade chaos can reduce the volume of double random phase encoding, improve the randomness of the sequence and keep the fastness of the Logistic chaos. It can also solve the problems of uneven distribution of the Logistic sequence. The HASH value SHA256 of plaintext is associated with the secret key strongly, which makes the whole system achieve the encryption effect of one image with one key. This also improves the avalanche effect between plaintext and ciphertext, and enhances the capability of resisting the attack of selecting plaintext and selecting ciphertext.

To improve the infrared small targets detection performance under complex background and noise interference, a single-stage infrared small target detection algorithm combining fully convolutional neural network and visual saliency is proposed. First, a lightweight fully convolutional neural network based on encoder-decoder architecture is designed to segment infrared images. The network can suppress the background and enhance targets simultaneously. Then, the saliency features of infrared small targets are used to further suppress false alarms. Finally, an adaptive threshold method is used to extract small targets. In the network structure, multiple subsampling layers are introduced to reduce computation load and increase the receptive field; multiscale features are introduced to improve the background suppression ability; attention mechanism is introduced to improve the training result of the model. Experiments on real infrared images show that the proposed algorithm is superior to the typical infrared small target detection algorithm with respect to detection rate, false alarm rate and computation time, and is suitable for infrared small target detection under complex background.

Aiming at the problem of low accuracy and poor real-time performance of commonly used target detection algorithms, a novel ship target detection algorithm based on convolutional neural network with feature fusions is proposed to detect multi-scale ship targets in complex scenes. The proposed method inherits the network structure of SSD and introduces the deconvolution feature fusion module and the pooling feature fusion module into it to generate the new feature maps with richer semantic information for both ship classification and boxes regression. In addition, we used a focal classification loss function in the training strategy to deal with the imbalanced difficult and easy samples in the training process. The experiments tested on the ship detection dataset demonstrate that the proposed method shows a better adaptability to ship detection of different sizes in complex scenes. On the extended experiment, the proposed method performance over SSD in blurry object detection.

As a solution of the slow space target acquisition caused by the complexity of starry sky background and the large field of view with relatively high sensitivity, a fast space target acquisition algorithm was proposed. This algorithm executes by taking advantage of star map recognition in large field of view. The speed of stars map recognition is largely improved by the short-time exposure on basis of the graded exposure strategy. Then, two-dimensional maximum background estimation method is used to realize star map preprocessing and obtain centroid data information. Finally, a natural object culling method based on star map target sequence is proposed. By horizontal culling of star map target sequence, the culling efficiency is effectively improved on the premise of ensuring the culling accuracy and realizing the rapid acquisition of space objects simultaneously. According to the simulation results, the proposed method is fast and accurate. Additionally, the period of target acquisition is reduced by 41.92% favorablely compared with competitors.

In the calibration process of the amplitude-division polarization imaging detector, a nonlinear least squares fitting method was proposed to solve the problem of strong coupling between the error introduced by the calibration system and the super-influencing factors of the detector''s own error. It substituted the main factors affecting the errors of the detector and the calibration system into the fitting equation in the form of unknown parameters for calibration calculations. The accuracy of the calibrated detector was verified under an incident light source with a wavelength of 532 nm. The polarization angle error of detecting linearly polarized light is below 0.2°, and the linear polarization error is below 2.5%, the circular polarization error of detecting left-right circularly polarized light is below 3.5%, which satisfies the application of sky polarization detection. It can be seen that the nonlinear least squares fitting method can effectively calibrate the amplitude-division polarization imaging detector.

A multi-spectral ghost imaging scheme was proposed to address the issues of image blurring and degraded signal-to-noise ratio occurring under the push-broom mode by superposing the detected signals. Taking the multi-spectral camera based on ghost imaging via sparsity constraints as the imaging system, in this scheme, adjacent frames of detected signals are shifted superposed, then the corresponding equivalent detection matrix is derived combining with the calibrated measurement matrix, and the image is reconstructed via compressed sensing algorithms. Simulation and experimental results show that:properly increasing the exposure time helps to improve the reconstruction quality; under the same exposure time, images reconstructed from the shift-superposed signals obviously have enhanced SNR than those from the original signals acquired via a single frame.

In order to satisfy the increasing calibration demand for high range luminance and improve the luminance calibration ability in China, a new generation of national luminance work standard was established which covers the range of 10~200 000 cd/m2. A twin light source injection adjustable high luminance uniform source was developed, and a precision limiting aperture was installed at the export of the source. The realization of the unitsc based on integrating sphere source method was introduced, and the key system characteristics were analyzed by experimental data, including the stability and spatial output characteristics of source, the position accuracy of adjustment, the linearity and the spectral mismatch error of the standard photometer etc. Finally the realized scale was transferred to a standard luminance meter which would be used as a transfer-reference. The relative expanded uncertainty of the new cd/m2 realization system is 0.7% (coverage factor is 2). Comparison between the proposed system and the current luminance work standard (3~1 500 cd/m2) shows a deviation less than 0.2%. The upper range of the proposed realization system is increased by 2 orders of magnitude compared with the original standard, and coveres the ranges of normal luminance meters, which can well satisfy the calibration demand in the high level luminance application of industry.

Aiming at the problem that the deformation measurement of thin-walled part is difficult and complicated, a method of deformation measurement of thin-walled part based on binocular vision is proposed. For the surface deformation of the part, a rigid metal block with reference coordinate system markers is installed on the measured part, and the designed coded targets and color circular markers are pasted on the surface of the part. The effective image region of the reference coordinate system and the coded targets are segmented by the color circular markers, which can eliminate the interferential image features. The identification and detection of coordinate system markers and coded targets are carried out, which realizes the accurate location of centers by using the designed corner structure, and then the 3D coordinates of the measurement points are calculated. The surface deformation of the part is calculated by comparing the changes of the 3D coordinates of the measurement points before and after part deformation. For the edge deformation of part, the improved Canny algorithm is used to extract the edge contour information of part, then the stereo matching and 3D reconstruction are carried out based on epipolar constraint and gray similarity. The experimental results show that the measurement method is reasonable and effective, and the measurement accuracy meets the requirements.

As a series of structural approximation need to be used in traditional geometry method, the model solving and accuracy analyzing of six-light-screen optical target based on this method are inaccurate. Here, a more accurate modeling and accuracy analyzing method based on plane equation was developed and a highly versatile engineering model and error transfer formula for six-light-screen optical target were deduced. Measurement accuracy decline caused by target distance and its error, light-screen tilt angle and its error, alignment error of light source and receiver, etc. were analyzed and compared under two kind of classical six-screen structural model. Sets of position and velocity measurement error distribution data and graphics in effective sensor area were acquired. A practical engineering layout of six-light-screen optical target with position measurement error less than 3 mm, and relative velocity measurement error less than 0.3% was proposed. This research can provide useful theoretical basis and data reference for practical design and accuracy estimation of six-light-screen optical target.

Multi-channel three-dimensional measurement based on phase calculation uses multiple optical channels at the same time, which results in the coupling of crosstalk, chromatic aberration, lens distortion and nonlinear response of devices, and affects measurement results and detection accuracy. Aiming at this problem, based on the error types of the collected images, the errors are divided into the light intensity error between the image pixel brightness to the actual brightness and the position deviation of spatial position distribution of image pixels. Then the light intensity error relation and the position deviation relation are constructed respectively and carry on the quantitative measurement to the detection system. Finally, a systematic error compensation method is proposed, which corrects the errors of intensity and pixel position of the collected image by using the established error relation, and compensates the influence of the above errors on the measurement results. The error compensation method proposed in this paper is used in multi-channel three-dimensional measurement of a standard step, the experimental results show that the measurement error can be reduced from 0.678 mm to 0.031 mm, and the method has practical value for the high-precision three-dimensional measurement of such a system.

NaYF4:Yb3+/Tb3+ and NaYF4:Yb3+/Tm3+ double doping fluoride nanomaterials with different doping concentrations were prepared by hydrothermal synthesis. In these materials, the Yb3+ was used as sensitizer to assist in luminous, Tb3+ and Tm3+ were added to the matrix sodium yttrium fluoride as activators of luminous center. The morphology and luminous properties of NaYF4:Yb3+/Tb3+ and NaYF4:Yb3+/Tm3+ nanoparticles were investigated by scanning electron microscope (SEM), X-ray diffraction andfluorescence spectrum. The X-ray diffraction patterns of the series of samples well match the NaYF4 standard card. Upconversion luminescence spectra of materials under 980 nm laser excitation were obtained and the mechanism was also analyzed. Excited by 980 nm laser, the emission spectra of NaYF4:Yb3+/Tb3+ including blue, green and red light, which correspond to the radiative transition of 5D4→7F6, 5D4→7F5, 5D4→7F1, respectively. The strong 480 nm emission can be seen in NaYF4:Yb3+/Tm3+ excited by 980 nm laser, corresponding to the electron transition energy band of 1G4→3H6. A strong red light emission band centered at 660 nm is corresponding to the 1G4→3F4 energy level transition.

The volume Bragg grating was prepared by using the "five-step" thermal fixing process, crystallization mechanism of the photo-thermo-rraphictive glass is described, and the ion migration process and NaF crystallization process are intuitively detected by Raman spectroscopy and scanning electron microscopy. The absorption band at 350~600 nm is found by transmission spectrum, and the colloid formed in the internal nucleation process is determined to be a silver bromide and silver nanocomposite.The key factors influencing the crystallized particles in photo-thermal-refractive glass were investigated from three aspects:nucleation process, NaF crystallization process and UV exposure dosage. It is found that the size of[Agn0·(AgBr)m] micelles is the key to determine the crystal size of NaF during nucleation. The crystallization process mainly affects the number of crystal particles and has little effect on changing the crystal size.As the exposure dosage increases, the crystallization size of NaF decreases first and then increases, the inflection point is around 1.8 J/cm2, and the crystalline particles are the smallest.

In order to suppress laser speckle, a metasurface structure with high transmittance and random phase function is designed, which utilizing the basic idea of time-averaged suppression speckle theory. Then integrated with MEMS technology, the design and manufacture of micro-rotation stages is realized, and to achieve a ultra-small speckle suppression device. The experimental results show that the laser speckle contrast can be reduced to 2.63%, which meets the needs of laser projection. This novel laser speckle suppression device is simple in use, high in energy utilization rate, low in cost, and easy to mass-produce.

In order to further improve the sensitivity and stability of aluminum-based plasmonic sensor, a plasmonic configuration named aluminum-graphene hybrid structure, was proposed based on phase modulation. Employing a high refractive index prism, the plasmonic configuration excited by the Kretschamnn mode, was designed by depositing graphene layers onto aluminum film. With the help of the transmission matrix theory, the variation of geometry parameters on sensing performance was studied when the excitation wavelength With the help of the transmission matrix theory, the variation of geometry parameters on sensing performance was studied when the excitation wavelength was set to 632.8 nm. The calculated results show that, the proposed aluminum-graphene hybrid structure designed by phase modulation have provided 2 orders of magnitude higher sensitivity compared with conventional surface plasmon resonance sensors designed by angular modulation. Moreover, the introduction of graphene can not only efficiently hinder the oxidation of plasmonic aluminum film, but also enhance the detection sensitivity as high as 83 times. For a tiny refractive index variation of 1.333~1.334 2 in sensing interface, the proposed configuration can provide a higher phase change of 94.663° and a phase detection sensitivity as high as 7.888 5×104 o/RIU. The proposed surface plasmon resonance configuration can provide a reference for designing low-cost and ultrasensitive plasmonic sensors.

In order to meet the need of high spatial resolution and high count rate in the imaging of far ultraviolet band, the method of acquiring the induced charge of the photon count imaging detector is described, which can be used in the imaging spectrometer for remote sensing of far ultraviolet band in the ionosphere of the earth. The relation between the parameters of the charge sensitive amplifier with pulse shaper is studied. The influence of the rise time of the charge sensitive amplifier and the time constant of the pulse shaper on the noise and counting rate is analyzed by numerical model. The selection principle of network parameters is proposed, and a front-end readout circuitry for photon count imaging detector based on wedge and strip anode is developed with discrete electronic components suitable for space applications. The experiment results show that when the shaping time is 500 ns, the equivalent input noise charge of readout circuits is about 230 e, and the rise time of the charge sensitive amplifier is better than 8 ns, which have better consistency with the calculated values. The spatial resolution of the measured images is up to 7.13 lp/mm, and the highest counting rate is 86.2 kcps, which meets the requirements of the far-ultraviolet band imaging spectrometer on-orbit mission operations.

In order to work out the measuring method of noise factor of microchannel plate, a method to measure the noise factor of microchannel plate by measuring the sensitivity of photocathode and signal to noise ratio of image intensifier is proposed. According to this method, the noise factor of microchannel plate is measured under different cathode voltage, microchannel plate voltage and anode voltage. The results show that the noise factor of microchannel plate change with the change of cathode voltage, microchannel plate voltage and anode voltage. The voltage of microchannel plate has the greatest influence on noise factor and the anode voltage has the least. When the voltage of microchannel plate is increased by 100 V, the noise factor increases by about 0.11. The noise factor increases about 3.3×10-4 for every 100 V of anode voltage. The increase of working voltage of microchannel plate means the increase of electron collision energy and the increase of secondary electron emission coefficient. According to the existing noise theory of microchannel plate, the noise factor of microchannel plate will be reduced, but the measured results are increased. The reason for this contradiction is that in the existing noise theory of microchannel plate, only the influence of secondary electron emission coefficient, detection rate and electron collision probability on the noise factor are considered, but the influence of collision electron energy on the noise factor is not considered, so the noise theory needs to be revised.

To solve the problem that the phase error of the optical phased array is difficult to calibrate, rotating element electric field vector calibration method is modified. Comparing with the traditional method, the new method modifies the algorithm and corrects the phase immediately after measuring the phase error of one unit, then measures and corrects the phase of the next unit. Possible π phase error in traditional method is avoided when large initial phase distribution and finite optical power measurement accuracy present. The simulation results show that the main lobe intensity calibrated with the original method reaches at most 70.8% of the ideal value on average, and the standard deviation is at least 19.1%. While after calibration with the modified method the main lobe intensity reaches at least 87.6% of the ideal value on average, the maximum standard deviation is 7.3%. The modified method produces statistically more accurate and predictable calibration result. A 9×9 optical phased array chip is manufactured. The initial grating lobe suppression ratio ofthe chip is 2.12 dB. Calibrated with this modified method the ratio reaches 4.68 dB. The effectiveness and practical application value of this method are proved.