As an artificial micro-structure that can flexibly control the light field at the sub-wavelength scale, metalens has received extensive attention in the field of optical imaging in recent years. In the design of the metalens imaging system, the metalens is usually designed only according to the focal length and working wavelength, without considering the imaging process, which often leads to poor imaging quality of the whole system. In order to improve the imaging quality, a joint optimization method is proposed to achieve high-resolution imaging based on metalens by incorporating the two processes of metalens design and image reconstruction algorithm into the design process of the imaging system. This method not only optimizes the parameters of the metalens to obtain a ideal point spread function, but also achieves high-quality image reconstruction up to 2 times the resolution of the input image.

Aiming at the problem of insufficient positioning accuracy of precision positioning platforms with large travel distances, a novel nanometer accuracy positioning system based on feedback of heterodyne laser interferometry with non-coaxial beams is proposed. The system consists of a composite feedback of heterodyne laser interferometry with non-coaxial beams and high-precision grating scale, which retains the advantages of high-resolution and large travel distance of high-precision grating scale, and combines with heterodyne laser interferometry with non-coaxial beams to form collaborative positioning which eliminates the influence of installation errors and deformation of grating scale on positioning accuracy, and improves the positioning accuracy under large travel distances. The system adopts predictive fine-tuning positioning technology based on BP neural network, which to some extent improves positioning efficiency. The experimental results show that within the travel range of 100mm, the system’s axis bi-directional positioning accuracy can reach 28nm and its axis repeated positioning accuracy can reach 26nm. Through predictive positioning, when the positioning end threshold is set to 20nm, the system’s positioning fine-tuning time can be shortened from within 1.65s to within 0.6s. At the speed of 5mm/s, the tracking error of the system can reach 100nm.

The satellite-Earth laser communication has the characteristics of long communication distance and complex transmission channel. In order to establish a stable and reliable satellite-Earth laser communication link, it is necessary to build a large-aperture ground station. The point position distribution and structure size of the flexible support of the 500mm large aperture primary mirror in the ground station are studied. This allows the primary mirror assembly not only to meet the stiffness requirements, but also to have flexibility to mitigate its own dynamic errors. The optimization process of optical, mechanical and thermal coupling is established on Isight platform.The accurate mirror RMS value is obtained by fitting the surface shape with Zernike polynomial. Taking the RMS value as the optimization objective, the optimal solution of the flexible support point position and the optimal solution of the dimension parameter of the flexible rod are obtained. After the optimization, the temperature range of the primary mirror component is extended by about 30℃ and when the pitch Angle changes, the RMS value of the primary mirror decreases by about 2nm.The main mirror shape meets the requirement of index less than 0.08λ.

Digital holographic imaging is a key technology to obtain wave front information of 3D objects, and obtaining high-quality holograms is the first condition. However, due to the constraints of the imaging sensor and the influence of the experimental environment, the obtained digital hologram is noisy and has low resolution, which affects the SNR in the holographic reconstruction. To overcome this constraint, a method based on deep learning is used to improve the hologram quality and signal-to-noise ratio. The results show that the algorithm can be used in the acquisition of multi-scale hologram, and obtain high quality hologram reconstruction effect is better, to reduce speckle noise, and compare the three kinds of loss function in the performance of network training.

The floating display is a very promising technology which can display images in the air, giving the viewers immersive and immediate experience. A floating 3D display system based on the integral imaging is proposed, which consists of an integral imaging 3D display, a semi-translucent semi-retroreflective mirror and a retro-reflective film. The working principle of the integral imaging 3D display and the floating display principle of retro-reflective film are analyzed. Therefore, combining the integral imaging 3D display technology with the floating display technology based on a retro-reflective film, the light rays from the integral imaging 3D display will reach the retro-reflective film through the reflection of the semi-transparent and semi-retroreflective mirror, and the retro-reflective film reflects the light at the angle of incidence and reconstructs the 3D image again, which solves the problem of depth inversion of the integral imaging 3D image while realizing the floating 3D image display. The system provides a new theoretical basis for floating 3D displays and a new method for solving the depth inversion of integral imaging 3D images.

The aperture depth of the diaphragm is designed to be 15 mm, the diaphragm aperture tension angle is designed to be 10° on the left and right, 6.5° on the top, and 3° on the bottom, and the diaphragm is installed behind a primary reflector, and the centerline of the diaphragm coincides with the corresponding main light. Through the specific aperture depth, tension angle and other structural design of the diaphragm so that the boundary inside and outside the amplitude of the distinction is obvious and the amplitude of the normal distribution, combined with the use of sensors will be light intensity and image source surface temperature in numerical display, easy to detect calibration. The design of the diaphragm is small in size, obvious detection effect, low cost, through the diaphragm and the logic of the sensor with the use of the car head-up display can be used for the detection of sunlight backflow, for the car head-up display industry to play a certain reference role.

Chemiluminescence combined with tomography (CT) can measure the three-dimensional information of the combustion flow field. In order to study the reconstruction effect of tomography, two different iterative algorithms including algebraic iteration (ART) and maximum likelihood expectation maximization (MLEM) are implemented. The reconstruction performance of the two algorithms is compared under different projection layout, different flow fields and different noise levels. The simulation results show that the non coplanar layout can effectively improve the reconstruction accuracy when the number of projections is small; When the number of projections reaches 9, the root mean square errors of ART algorithm and MLEM algorithm are both less than 2%, and MLEM algorithm is slightly better; MLEM algorithm has stronger anti-noise ability and is more suitable for reconstruction in complex experiment environment; However, the time cost of ART algorithm is superior. ART algorithm is 56.57% faster than the MLEM algorithm for 9 projections. The experimental results show that both algorithms give good reconstruction results, and the MLEM algorithm has higher computational accuracy than the ART algorithm. With more voxels divided in the flow field to be measured, the time cost consumed by MLEM is about 8.37 times higher than that of the ART algorithm.

Digital holography has the advantages of high measurement accuracy and real time, and has high application value in many fields; photopolymer, as an optical storage medium, is one of the most promising holographic recording materials, but the thickness shrinkage caused by the photopolymerization reaction during the recording process greatly affects the recording effect, and in order to measure the specific shrinkage of its thickness, a digital holography-based measurement method for the shrinkage of photopolymer is proposed. The experimental system is designed based on the Mach-Zender interferometric optical path, and the holograms of the photopolymer before and after the reaction are recorded, and the phase change is calculated by using the holographic interferometry method.

In 3D shape measurement technology, phase demodulation is a key step to analyze deformed fringe patterns to get object surface information. Phase shift method is a non frequency domain analysis method, which requires three or more fringe patterns and is only applicable to static measurement. However, phase demodulation methods based on frequency domain analysis, such as Fourier Transform, Windowed Fourier Transform, Wavelet Transform and Hilbert-Huang Transform, only need one fringe pattern to extract phase, which are easier to operate and more automatic, and suitable for dynamic phase extraction. The status of several phase demodulation methods are researched, and the basic principle of the phase demodulation methods and its demodulation effect on the phase are deeply analyzed, and finally summarizes the characteristics of the phase demodulation method based on frequency domain analysis and its prospects for the future.

A three grating monochromator with a wavelength range of 185~900nm was designed to meet the accuracy requirements of UV visible spectrum calibration; To improve the wavelength scanning accuracy of monochromators and reduce the engineering cost of high-precision wavelength scanning mechanisms. A model was established to investigate the impact of grating angle error on the wavelength accuracy of the scanning mechanism. A high-precision encoder based angle error compensation device was designed, and the accuracy correction of the wavelength scanning mechanism was achieved through compensation algorithms designed with different software parameter settings and algorithms; The experimental results show that after compensation through parameter algorithms, the wavelength accuracy and wavelength repeatability accuracy of the scanning mechanism have been improved within the working wavelength range. The degree of accuracy improvement varies under different wavelengths, with an improvement ratio of 20.8% to 35.2% for wavelength accuracy and 61.5% to 95.2% for wavelength repeatability accuracy. The above experimental results verify the effectiveness of the error compensation device and compensation algorithm, improve the accuracy of the wavelength scanning mechanism, and reduce the cost of high-precision wavelength scanning mechanisms in engineering applications.

Aiming at the problem that the positioning precision of received signal strength based indoors positioning methods for visible light communication system is low, a new indoors positioning method for visible light communication system based on deep neural networks is proposed. In this method, the visible light channel estimation technique is adopted to measure the indoors distance, so that the problems of insufficient stability and reliability of the received signal strength are resolved. Besides, a deep neural network is designed to learn the distribution characteristics of the distance vectors of the photodiode in offline phase, in order to avoid the problem that the instable light signals lead to error growth. In online phase, the target is positioned based on multiple distance vectors, thus the positioning precision can be improved further, at the same time, the time efficiency meets the requirements. Simulation results show that, in the indoors scenario, the proposed method achieves better positioning precision than traditional triangulation methods and received signal strength based positioning methods.

Based on the photorefractive media, the propagation properties of hollow sinh-Gaussian beams are investigated by using the finite difference method. The intensity envelopes of hollow sinh-Gaussian beams with different beam orders are obtained, and the hollow region becomes larger as the increasing of beam orders. The influence of beam orders and nonlinear parameters on the propagation pattern of beams is analyzed. When the beam order is 2, the hollow sinh-Gaussian beam presents the periodic focus-behavior and the focusing intensity ratio gradually becomes weaker within a certain transmission distance. With the increasing of nonlinear parameters, the first focusing distance decreases gradually, but the focusing intensity ratio increases. When the beam order is 3, the focus-behavior of beams presents only once and the beam shows spotty pattern. When the beam order is 5, the first focusing distance and the focusing intensity ratio increase further. All the results show that the propagation properties of hollow sinh-Gaussian beams can be controlled flexibly by adjusting beam orders and nonlinear parameters.

Because of its large area, heavy weight, processing quality is difficult to guarantee, and the production efficiency is extremely low, which seriously limits the development of such products. Consulting the domestic and foreign scientific and technical literature, The research on processing technology of large-aperture optical lens is seldom involved.At present, the commonly used process is still the traditional processing method, the production quality is difficult to guarantee, and the production efficiency is extremely low. High-speed processing technology of spherical optical lens with diameter of something 200mm ~ something 300mm process adopts the process of fixing the lens on the main shaft to carry out the active rotation movement, and the mold is on the top, and the mold moves according to the rotation of the lens. In order to obtain a stable surface profile, the principle of surface forming was analyzed, and the technological factors in the actual production process were studied. Processing of large diameter spherical optical lens, Milling adopts vertical milling machine for processing, The processing time is about 30~60s; The fine grinding is processed by fixed abrasive diamond pellets, Divided into two sand W14 and W7 matching, The processing time is about 30s and 20s respectively; The polishing mould is made of polyurethane polishing sheet. The processing time is about 300s.The processed lens is tested for the shape,PV is 0.227λ, RMS is 0.024λ. And the ideal processing effect is achieved.

Firstly, the application background and advantages of star map recognition algorithm are introduced. The three mainstream star map recognition algorithms, which are the subgraph isomorphic algorithms, pattern recognition algorithms and artificial intelligence algorithms, are introduced, analyzed and compared. After that, the evaluation of star map recognition algorithm research is explained. Finally, the future development of star identification algorithm is prospected, including the future research on star identification in non visible light, star identification in high dynamic complex environment and star identification considering aero optical effects.

Automatic segmentation of retinal vessels facilitates the diagnosis and treatment of retina-related diseases. The task is challenging because of the complex structure of retinal vessels and interferences in fundus images, such as low contrast, uneven illumination, and pathological exudates. To address the lack of global semantic dependency modeling and the semantic gap between encoders and decoders in the U-Net for this task, an Intra- and inter-scale augmented U-Net (I2A-Net) is proposed. I2A-Net is designed based on two perspectives: for the intra-scale encoding-decoding layer, a Spatial Enhanced Self-attention mechanism is integrated in each coding layer to enhance the global spatial aggregation capability, and further developed into the decoder to alleviate the information loss caused by up-sampling operations; For inter-scale encoding-decoding layer, a novel Cross-scale Fusion module is introduced to boost the semantic interaction with other layer by dynamically leveraging the rich feature information of the deepest layer, thus further alleviating the semantic gap between encoder and decoder. The experiments on three retinal standard datasets, DRIVE, CHASE~~DB1 and STARE, demonstrate that I2A-Net can effectively segment the retinal vessel structure, and compared with Baseline, I2A-Net can yield superior performance in all evaluation metrics.

Single molecule immune detection refers to limiting immune complexes to a very small volume and counting the generated signals. It is a "digital" immune detection technology. The core of the single molecule immune detector is similar to a fluorescence microscope system, but the common fluorescence microscope has a complex structure and cannot match the specification of the biochip, which cannot meet the detection requirements well. A set of high performance optical system for single molecule immune detection is summarized according to the specification of biochip developed by ourselves. Firstly, the appropriate initial structure is determined according to the design requirements; Secondly, on the basis of fluorescence microscope system principle and aberration analysis theory, the optical design software Zemax is used to repeatedly optimize the system design; Finally, an optical system is designed to meet the imaging requirements of single molecule immune detection. The imaging part of the system consists of 15 pieces of refractive lenses, the working object distance is 4mm, the magnification is -4.52, the modulation transfer function values are greater than 0.44 at the Nyquist frequency of 73lp/mm, and the distortion is 0.8%, Kohler lighting mode is adopted for the lighting part, and the illuminance uniformity on the object surface reaches 97% in the whole field of view. The whole system adopts domestic conventional glass, which is conducive to processing and cost reduction. The optimized high resolution fluorescence microscopy system combined with the current fully automated detection technology can greatly improve the detection sensitivity and accuracy of single molecule immunoassay.

As a Digital PCR (dPCR) technology, an updated iteration of traditional PCR technology with absolute quantification, has great potential in biological fields such as neo-coronavirus nucleic acid sequence detection and cancer cell detection. Existing dPCR fluorescence detection systems are challenging to make a reasonable trade-off between detection throughput, detection speed, and cost. The small imaging field of view and multiple image stitching greatly limit the accuracy and detection time of dPCR fluorescence detection systems. Therefore, a digital PCR detection experimental platform is builded with multi-channel based on a large field-of-view fluorescence microscope system. Rely on this platform,By simulation experiments of Zemax to improve the design of microscopy objective, Relying on this experimental platform, Zemax software is used to optimize the system structure and improve the design of microscopic objectives, which can realize an 18mm full field of view and 6.3μm resolution among the FAM、HEX、ROX channels. By using the Randomized Hough transform algorithm (RHT), the microfluidics chip including the normal "holewell" structure、the new structure such as "snowflake", "branch" and other Irregular structures can be obtain a clear and stable fluorescence images under the premise of a large field of view and fast detection.

Due to the influence of processing, installation and adjustment of nasal endoscopes in China, the resolution and aberration of endoscopes in actual use are significantly worse than the design level. Through tolerance analysis, it is concluded that the eccentric tolerance has a great impact on the results. Based on the above problems, the installation and adjustment error on the basis of a nasal sinus objective are simulated with a field angle of view of 80 ° and an entrance pupil diameter of 0.2mm, It is proposed that a phase plate is added in front of the protective glass of the objective lens to compensate for the decline of imaging quality caused by the assembly and adjustment error without changing the mechanical structure of the post assembly and adjustment nasal sinus objective lens; It can be seen from the design results that after adding the phase plate, the system transmission diagram increases from 30LP/mm equal to 0 to 170LP/mm greater than 0.2; Using N3 discrimination plate to detect the system resolution, about 18 groups of objects can be distinguished, and the object resolution is 15μm. It can be seen that the design based on the compensation phase plate can significantly improve the image quality without changing the original design.