In order to study the changing rules of the reflective characteristics of the reflective film in a curved state， the mapping relationship between the dihedral angle of the reflective surface of the curved Cube Corner Reflector （CCR） and the bending angle is established based on the principle of volume invariance. The reflection theory analyzes the reflection characteristics of a single curved CCR and a curved reflective film， and obtains the axis-symmetric distribution of the light rays emitted by the single curved CCR and the maximum theoretical effective bending angle of the curved reflective film is 70.52°. Finally， using ray tracing software to simulate a single curved CCR and curved reflective film， the results show that the diffraction pattern of a single curved CCR and curved reflective film is axisymmetrically distributed when light is normally incident； the diffraction pattern of a single curved CCR and curved reflective film is axisymmetrically distributed； the maximum effective bending angle of the curved reflective film decreases with the increase of the bending angle， a decrease of 0.7%， and the deviation between the value and the theoretical value is small； when the bending angle is greater than 20°， the reflectivity of the convex reflective film is increased by 20% compared with the concave reflective film. At the same time， the experimental measurement results prove the correctness of the theoretical analysis.

Based on the tightly focusing approach， the perfect optical vortex beams are generated at the geometric focal plane. The optical forces and torques exerted on the micro-sized particle in the perfect optical vortex beams are calculated. Numerical results show that the micro-sized particle is trapped on the ring of perfect optical vortex beams and is driven to rotate around the optical axis. The orbital torque will firstly increases and then tends to be stable with the increase of the topological charge value. In addition， the influence of circular polarization state， radial polarization state and azimuthal polarization state on the force and orbital torque of micro-sized particles are investigated in perfect optical vortex beams. The results show that the state of polarization of perfect optical vortex beams will affect the optically induced orbital motion to some extent. The circular polarization state of perfect optical vortex beams is superior to drive the orbital motion of the particle.

A fiber bundle made of 12×9 individual Ge-As-Te-Se fibers with a core diameter of 20 μm has been tested for optical properties and used to carry out the imaging experiment. A 5~11 μm continuous tunable quantum cascade laser has been used to test the bundle’s attenuation property with a cut-back method. The attenuation of bundle is nearly 1 dB/cm in the spectral region. A compact distal objective lens with length of 13.6 mm and diameter of 6 mm based on telecentric system has been designed and fabricated. The MTF of the lens showed the ablitity to transmit the image with resolution of 25 μm. A infrared image with 2 mm×2 mm and 100 μm resolution with quantum cascade lasers illumination， which has high power spectrum density and narrow linewidth， has been captured. A comparison experiment about the environmental temperature has been conducted illuminated by quantum cascade laser and incoherent black body， it proved that the quantum cascade laser illumination could help build a high SNR imaging under a high environmental temperature even above human's temperature. It would be meaningful for the next infrared endoscopy experiment in vivo way.

In order to achieve high sensitivity space laser communication and improve the anti-interference ability of the transmission channel， this article combines single photon detection technology and pulse position modulation technology， the single photon detector avalanche is quenched by the combination of the gated circuit and the feedback quenching circuit. The insert frame header method was designed to modulate and demodulate the pulse position. The pulse position modulation and demodulation process was simulated by field programmable gate array， which verified the effectiveness and feasibility of the insert frame header method. On this basis， a 1 550 nm pulse position modulation laser communication experiment was built， and the performance of the single photon detector under different parameter was tested at the same time. The results show that the single photon detector has the best performance when the detection efficiency is 25%， the trigger delay is 8.00 ns， the gate width is 5.0 ns and the death time is 0.1 μs. Finally， the detection sensitivity of single photon detectors with different modulation rates was tested. The results show that the communication sensitivity is -51.8 dBm when the communication code rate is 1 Mbps； the communication sensitivity is -41.0 dBm when the communication code rate is 4 Mbps， which realizes high sensitivity space laser communication.

To improve on the poor real-time performance and low accuracy of traditional manual feature extraction and pattern recognition method in the Phase-Sensitive Optical Time-Domain Reflectometric System （φ-OTDR）， a new pattern recognition method based on wavenet is proposed. This method fully analyzes the temporal causality of optical fiber vibration signals through the causal dilated convolutions， and makes the model converge faster by the residual block structure， so as to achieve higher recognition accuracy and efficiency.The experimental results show that， when three signals are recognized， namely， hand tapping， foot stepping， and stick striking， compared with the other two common methods， one-dimensional convolutional neural network structure and long short-term memory structure， the recognition accuracy as high as 99.85% is achieved with the proposed method. And it consumes the least amount of training time， as short as 96 s. Also， its signal detection process takes only 30 ms， which can meet the real-time application's requirement. The proposed pattern recognition method has high accuracy and good real-time performance. It should be of great significance for the application and popularization of φ-OTDR systems in perimeter security and similar areas.

As conventional hyperspectral image classification algorithms can not solve the problem of spectral deviation in different images well， a hyperspectral image classification algorithm based on dense convolution and domain adaptive is proposed. First， dense convolution is used in the source domain to perform deep feature learning， and then apply domain adaptive technology to transfer to the target domain. Convolutional neural networks are commonly used for feature learning in the current domain adaptive hyperspectral image classification framework， but when the depth increases， the classification accuracy may decrease due to the disappearance of the gradient. Therefore， this paper introduces dense convolution for deep feature learning， to improve the accuracy of domain adaptive hyperspectral image classification. The effectiveness of the proposed algorithm is verified on the Indiana hyperspectral dataset and Pavia hyperspectral dataset. The overall classification accuracy is 61.06% and 89.63%. Compared with other domain adaptive hyperspectral image classification methods， the proposed method has better classification accuracy.

Aiming at the varying scene light in imaging process and difference between the haze relevant features in image dehazing，a channel attention network and fuzzy partition entropy with graph cut for single image dehazing based on the improved atmospheric scattering model with varying scene light is proposed. Firstly， the encoder-decoder network with channel attention mechanism is utilized to estimate transmission map. Then the proposed channel attention module is applied in the end of encoder and the beginning of the decoder for assigning different weights to different haze relevant feature maps and obtaining accurate transmission map. Then， the fuzzy partition entropy combined with graph cut is used to segment the transmission map into distant scene， middle scene and close scene covered with varying scene light. This scheme combines spatial correlation and fuzzy partition entropy， solving misclassified problem introduced by the threshold-based segmentation. Finally， a clear image is obtained with the predicted transmission map， estimated scene light and atmospheric light. Extensive experiments demonstrate that this method achieves promising effective on synthetic images and real images. Comparing with exiting methods， our method improves dehazing results in both peak signal to noise ratio and structural similarity. The average running time for handling single hazy image is 3.9 s.

Aiming at the problem of information redundancy in the pairwise synthesis of heterogeneous difference features during the fusion of dual-mode infrared images， which leads to conflicts between the selected fusion algorithms， resulting in poor fusion or even failure， a method of dual-mode infrared image fusion algorithm selection based on possibility information quality synthesis is proposed. Firstly， the fusion effectiveness of different difference features under the dual-mode infrared image multi-fusion algorithms is calculated， and the possibility framework is used to obtain the corresponding possibility distribution vector subsets. Secondly， the information quantity and credibility of the vector subsets is calculated， and multiple vector subsets are weighted and synthesized. Then a ranking function based on information quality is constructed to obtain the non-dominated subsets under each fusion algorithm. Finally， the joint distribution of multiple fusion algorithms score function is constructed to optimize the selection of multiple fusion algorithms. The experimental results show that the method of integrating multiple difference features based on quality is used in the selection of dual-mode infrared image fusion algorithm， the selected fusion algorithm is more than 55% higher than the average of other algorithms in the weighted comprehensive index， which proves the effectiveness and feasibility of the method. The average time of the proposed method is 10.083 s after several experiments， which also meets the engineering requirements of real-time image fusion application in terms of time efficiency.

In order to improve the visibility of fused images and solve the problems of missing edge features and fuzzy details in traditional infrared and visible image fusion algorithms， an novel image fusion algorithm in latent low rank representation framework based on convolution neural network and guided filtering was proposed. Firstly， the source images are decomposed to low-rank parts and saliency parts by latend low-rank representation. Secondly， according to the pixel activity information of source images， the weight maps are obtained through the convolution neural network. Thirdly， the weight maps are improved by guided filtering according to the source images and through that the weight maps of low-rank parts and saliency parts can be obtained respectively. Then， the weight maps are fused with low-rank parts and saliency parts of original images to obtain the low-rank part and the saliency part of fused image. Finally， the final fused image can be obtained by adding the fused low-rank part and the fused significant part. Compared with other fusion algorithms， the experimental result shows that the proposed algorithm is superior to the traditional infrared and visible image fusion algorithms in terms of subjective visual effects and objective indexes.

In the lidar long-distance imaging scene with high background noise and low integration time， in view of the problem of the depth image target obtained by traditional methods being submerged by noise and the large deviation of depth estimation， a method based on signal photon time correlation and adaptive Kalman filter depth information estimation method is proposed. The photon counts with aggregation feature in time will be extracted to form a set firstly； then the factors that affect the temporal distribution of signal photons will be analyzed and Gaussian linear model will be used to described the photon set； finally， the time-of-flight of all photons in the set will be scrambled and input into the improved adaptive Kalman filter to iteratively estimate the depth value. In the room with signal to noise ratio of 1， compared with the traditional maximum likelihood method， this method improves the root mean square error by 40% and 38% when the integration time is 10 ms and 1ms respectively. In the outdoor 2 km target imaging experiment with signal to noise ratio of about 0.135， when the signal photon numbers are 100， 33 and 17 respectively， the depth image of this method is clearer and the noise is lower than the traditional maximum likelihood method and fast denoising algorithm with the temporal correlation of photons. It is verified that the method can be applied to the depth information estimation and image restoration of lidar remote imaging under high noise and short integration time.

An auto focusing method based on simulated annealing algorithm is presented to recover the respective high resolution images for three different illumination colors， thus to recover the high resolution color image. Simulation modeling as well as the experiments on resolution target and biological slices all demonstrate that the high resolution color image can be recovered even for the microscopic system with serious chromatic aberration， which significantly relaxes the request for high quality microscope， thus expanding the application scope of the Fourier ptychographic microscopy.

Single-pixel imaging has made significant progress in imaging quality and speed. However， how to improve the imaging speed while ensuring the imaging quality is still a problem to be solved. Based on the finding that the average value of the absolute value of the Hadamard coefficients of each oblique line declines rapidly， an adaptive oblique zigzag sampling method is proposed to achieve the expected imaging speed and quality. Experimental results show that the number of sampling and the amount of data acquisition can be effectively reduced using the proposed method， improving the imaging speed while ensuring the imaging quality. Compared with the single-pixel imaging using sequential sampling， the imaging quality of the proposed method is better under the same sampling rate.

In industrial applications， only one side of the composite media can be detected in most cases， so it is necessary to study the terahertz nondestructive testing method to detect the internal parameters of the desired media based on the reflection mode. The transmission model of terahertz waves in the medium of a three-layer structure is derived under the reflection mode. The thickness of the middle layer together and its refractive index are estimated by a genetic algorithm to attain the specific information about the middle hidden layer. At the same time， samples with the three-layer structure consisting of the hidden layer with the thickness of 200 µm are prepared and subjected to a terahertz time-domain spectroscopic system. In addition， a theoretical model is designed to compare the achieved measurements. Thereafter， a genetic algorithm is designed to estimate the thickness of the hidden layer and the respective refractive index to verify the effectiveness of the proposed method. Compared the estimated value of thickness with the measurement result of the thickness gauge， the error is kept within 4%. Compared with the actual value， the error range of the estimate refractive index fluctuates greatly， the average error is about 6%. Finally， the error sources is analysed， which provides the theoretical and experimental basis for the iternal defects of the multilayer composite material and the dielectric parameter of the intermediate layer material. The practice shows that the system as a nondestructive evaluation method can be widely used in the reliability evaluation of layered structures.

Aiming at the problems of reduced navigation star points positioning accuracy and increased attitude error caused by dynamic star points imaging shape distortion and imaging position error of the star tracker in the rolling exposure mode， a navigation star points correction method based on time domain constraints is proposed. Based on the time asynchronous characteristics of rolling shutter exposure， combined with the optimal estimation of navigation star points positions and inter-frame velocity based on Kalman filter， the navigation star points at different moments in the time domain are corrected to the same time， thereby effectively solving the problem of star points distortion and position offset. The simulation experiments under different conditions were carried out. The experimental results show that when the angular velocity of the star sensor is within the dynamic range of 0~3°/s， the distance accuracy of the star pair after correction is higher than 0.1 pixels， attitude angle accuracy is higher than 10".

For the purpose of effective analysis on rear irradiance of multi-row bifacial photovoltaic module arrays fixed with constant slope angle， a key operation for modifying back side irradiance model of bifacial photovoltaic modules in existence is that a distance of D is divided into n segments with distinguishing shaded and unshaded regions. Subsequently， modified model is achieved by optimizing view factors of diffusion irradiance and reflective irradiance. The simulation and experiments results show that the total radiation of bifacial photovoltaic modules obtained a gain of 18.98% relative to signal photovoltaic modules in the condition of a clear weather， together with the slope angle of 26 °， the height from the ground of 1.5 m， the distance between rows of 4 m， the length of the module of 1.7 m， and the reflectivity of the ground of 30%. Moreover， the irradiance gain error could be controlled only 0.68% in the condition of aforementioned.

A passively Q-switched state or a pulse amplitude chaos output state are experimentally realized based on a diode-pumped Nd：LaMgAl11O19 disorder crystal laser. Experimental results indicate that when the pump power is in the range of 4.8~8.6 W， the Nd：LMA laser operates at a Q-switched state. For a pump power of 8.6 W， the average output power， repetition frequency， pulse width are about 613 mW， 157.1 kHz， and 2.2 μs， respectively. When the pump power is in the range of 8.7~10.5 W， the pulse amplitudes of the Nd：LMA laser present irregular and random distributions， which are judged as chaos through analyzing the autocorrelation curve， phase portrait， power spectrum， and stochastic histogram of the pulse-peak time series. For a pump power of 10.5 W， the average output power arrives at about 814 mW.

A compact pulse side-pumped Nd： YAG laser by high-temperature laser diode arrays was designed. The operating temperature of pumping source was controlled at 60℃ by the semiconductor refrigerator， the emission center wavelength was 808nm， and the spectral line width was 4.12 nm. Dynamic temperature field distributions of the pumping source in 60s at 40℃、50℃ and 60℃ were simulated. Two φ5 mm×50 mm Nd：YAG crystals with 1.0at.% Nd3+ doping concentration were used as the laser gain medium. The crystal of potassium dideuterium phosphate was used forelectro-optic Q-switching. With a pump source voltage pulse width of 250 μs and the repetition frequency of 20 Hz and 1 Hz， output pulsed energy of 230 mJ and 246 mJ were achieved corresponding to the pulse durations of 8.4 ns and 7.8 ns. The beam divergence was 1.6 mrad. The total electro-optical conversion efficiency of designed Nd： YAG Q-switchedlaser was greater than 4.6%.

A Flexible Liquid Crystal Micro-Lens Arrays （LC MLAs） with an adjustable focal length was designed and prepared to solve the narrow viewing angle of the integrated imaging. In this paper， the hole-patterned driven- electrode arrays were fabricated by a photolithography technology and the flexible Polyimide （PI） film was prepared by a spin coating method on the flexible ITO glass substrate. After the PI film layer was heated at 60°C for 5 min， the PI orientation layer was formed by using the plasma at a power of 630 W for 5 minutes. The optical properties of LC MLAs with a flat substrate and a curvature radius of 7.5cm were studied after the LC MLAs were fabricated. The experimental results show that the prepared LC MLAs can achieve the excellent focusing properties under the condition of a flat plate and a curvature radius of 7.5 cm. The interference patterns are uniform and the focal radius are small when the operating voltages are applied to the LC MLAs. Further， the focal length of the flexible LC MLAs with the curvature radius of 7.5 cm can be tunable from 0.43 mm to 1.05 mm when the operating voltages are changed from 3 Vrms to 5.3 Vrms.

In order to overcome the shortcomings in obtaining white light emission due to the anion exchange reaction from different halide perovskite quantum dots and unstable red light emission halide perovskite quantum dots， a method was used to prepare rare earth ions Tb3+/Eu3+ co-doping into perovskite quantum dots in ambient air. By adjusting the doping ratio of Tb3+/Eu3+ rare earth ions， the energy transfer from the host lattice of perovskite quantum dots to Tb3+/Eu3+ was accordingly manipulated to obtain single-component and white light emission perovskite quantum dots of （Tb， Eu）∶CsPbCl3 and （Tb，Eu）∶CsPb（Cl/Br）3. Meanwhile， we delicately investigated the morphology， structure， luminescence performance， energy transfer mechanism and stability of quantum dots. The results show that under 365 nm laser excitation， white light emission with 1 931 Commission Internationale de l’Eclairage color coordinates was achieved in perovskite quantum dots （Tb， Eu）：CsPbCl3 with different contents of Tb3+/Eu3+ ions. When the feed ratio PbCl2∶TbCl3∶EuCl3 is 1∶1.5∶1， the quantum yield is 3.59%， which is 6 times higher than the quantum yield （0.57%） of pure CsPbCl3 quantum dots. Further research found that the （Tb， Eu）∶CsPbCl3 quantum dots were stored in the air for 2 months， the quantum yield was almost unchanged （3.63%）， maintaining good stability. Moreover， the luminescence characteristics of the co-doped perovskite quantum dots with Tb3+ and Eu3+ synthesized with different solvents （n-octane， 1-octadecene） were compared. Tb3+/Eu3+ co-doped perovskite quantum dots （Tb， Eu）： CsPbCl3 ware used to achieve single-component white light emission with good stability and has certain application prospects.

Studies on the passivation of monocrystalline silicon wafers by using plasma enhanced chemical vapor deposition for the deposition of monolayer intrinsic hydrogenated amorphous silicon films show that increasing the hydrogen dilution ratio is beneficial to reducing the defects in the films and enhancing the passivation effect. Excessive hydrogen dilution ratio can lead to the epitaxial growth of amorphous silicon on the silicon wafer surface and reduce the passivation effect. Annealing results in the increase of the degree of crystallization of amorphous silicon and the reduce of passivation effect. Meanwhile， annealing improves the quality of thin films and changes the way of H bonding and enhances the passivation effect. Therefore， the best passivation effect of monolayer hydrogenated amorphous silicon can be obtained only at appropriate hydrogen dilution ratio and annealing temperature. In order to improve the passivation effect of amorphous silicon film on silicon wafer， tandem intrinsic amorphous silicon film with high and low hydrogen dilution ratio is used to passivate silicon wafer. Therefore， the epitaxial growth of amorphous silicon on the silicon wafer surface can be avoided by stacking amorphous silicon thin films with high hydrogen dilution ratio on the thin films with low hydrogen dilution ratio. In the annealing process， the hydrogen in the film with high hydrogen dilution ratio diffuses into the film with low hydrogen dilution ratio， effectively passivating the suspension bonds on the surface of amorphous silicon and monocrystalline silicon， and improving the interface quality of amorphous silicon/silicon wafer. After the tandem passivation of the silicon wafer， the effective minority carrier lifetime is 7.36 ms， and the implied open-circuit voltage is 732 mV.

WO3 nano-rods were synthesized on the FTO substrate by hydrothermal method. Pt nanoparticles with different deposition time（40 s， 80 s， 120 s） was loaded onto WO3 nano-rods by an electrodeposition method to prepare WO3/Pt composite film photoanodes. Characterization of samples was conducted by scanning electron microscopy and X-ray diffraction， the results show that the WO3/Pt composite films have been synthesized. Diffuse reflectance spectra show that the WO3/Pt composite film has more strong absorption than pure WO3 nano-rods film. Electrochemical impedance spectroscopy shows that the WO3/Pt composite film enhances a charge transfer efficiency compared with pure WO3 nano-rods film. The photoelectric properties of the samples were obtained from the photocurrent and the photoelectric catalysis. The WO3/Pt composite film samples have higher photocurrent and photoelectric catalytic （PEC） activity than pure WO3， and the sample obtained by depositing for Pt nanoparticles at 80 s （WO3/Pt-80 s） has the highest photocurrent and photoelectric photoelectric catalytic activity. Meanwhile， the photoelectric catalytic activity of WO3/Pt-80s composite film is higher than direct photocatalysis or electric catalysis of WO3/Pt-80s composite film.