This PDF file contains the front matter associated with SPIE Proceedings Volume 12069, including the Title Page, Copyright information, and Table of Contents

The primary mirror component is an important part of the Cassegrain system. As the first-stage imaging component, the RMS surface error directly affects the image quality of the whole optical system. In this article, taking the primary mirror component of a certain type of Cassegrain aerial camera as the research object, the factors affecting the RMS precision of the primary mirror surface are analyzed in detail from aspects of back supporting structure design, platen elastic crimping design, simulation analysis, test verification and so on. Using the finite element method to simulate the primary mirror supporting structure, analyzes the influence on the primary surface error by the three-point supporting structure in different positions. Furthermore, analyzes the variations of the primary mirror surface error under the influence of three-point supporting structure and pressure plate. The last but not the least, analyzes the primary mirror surface error under the different pressure conditions, concludes the optimal supporting point position and the excellent elastic compression. After the primary mirror assembling, through test verification, the RMS is 0.0270λ, which is better than the original design requirement of λ/35(0.0286λ). And the RMS variation between before and after assembling is less than 0.005λ. Performing the high and low temperature test on the primary component, after test, the RMS values is 0.0269λ, it proves that the primary frame structure and its axial supporting structure have little effects on the RMS precision of the primary mirror. It can also meet the requirement of the large-aperture primary mirror surface in the co-optical system under complex conditions. The feasibility of the structure design has been verified.

6s(Second Simulation of the Satellite Signal in the Solar Spectrum) radiative transfer model is one of the atmospheric correction algorithms based on the atmospheric radiative transmission model. It is widely used because of its high correction accuracy. Meanwhile, it is criticized for the complexity of the parameters and the efficiency of the correction process. 6S model needs to establish a look-up table based on the geometric conditions and aerosol conditions which directly determines the accuracy of the atmospheric correction. This paper analyzes the limitations of traditional look-up table method and uses artificial intelligence algorithms such as the support vector regression(SVR) algorithm and the back propagation (BP) algorithm to instead the traditional look-up table method. The experiments’ results show that the output value and predictive value fit well. Both are better than the traditional linear interpolation performance results, and the BP algorithms performs better, which verifies the feasibility of BP neural network algorithms prediction model instead of linear interpolation method for table lookup. Finally, this paper takes Landsat-8 data as an example, uses the method proposed in this article to perform atmospheric correction, and compares the FLAASH model correction results. The visual performance results of the two are roughly the same.

We report on high-efficiency visible and near-infrared transmission gratings in fused silica generated by holographic recording and reactive ion beam etching technology. At a wavelength of 740 nm, near 100% diffraction efficiency is achieved under Littrow conditions. The design is based on the phenomenon of the high aspect ratio gratings by using the rigorous coupled wave analysis. A binary grating with the optimum grating period of 740 nm and groove depth of 1.55 had been fabricated in the paper. The grating wavelength bandwidth and angular bandwidth are extremely enhanced compared with conventional volume phase holographic gratings, making these gratings the key elements in high-resolution astronomical ground-based telescope spectrographs.

A kind of swing micro-mirror structure for gravitational wave observatory in space is presented in this paper. Harmonic response analysis and random vibration analysis are carried out. The analysis results show that the mechanism is able to bear the effect of the load during launch and has high dynamic stiffness.

As a new generation of optical gyroscope, FOG has been widely used in many important fields. With the wide application of high-precision FOG, users put forward higher requirements for the reliability of FOG. As a new research hotspot, the remaining life prediction and evaluation of high-precision FOG has become the focus of many technicians. This paper attempts to combine the residual life evaluation of high-precision FOG with deep learning algorithm, and uses deep learning method to evaluate the residual life of high-precision FOG. Experiments show that the method can effectively predict and evaluate the residual life of high-precision FOG. It achieves the purpose of accurate maintenance, repair and replacement, and is of great significance to improve the reliability of high-precision FOG.

Advanced microscopy techniques have opened new opportunities for biomedical research. Fluorescence microscopy enables researchers to observe subcellular structures with specific labeling. Quantitatively measuring the dynamics of intracellular objects is essential to understand the underlying regulatory mechanism. Protein-containing vesicles in cell are involved in various biological processes, such as material transportation, organelle interaction and hormonal regulation, whose dynamic characteristics are significant to disease diagnosis and drug screening. Although there have been some algorithms developed for vesicle tracking, most of them have limited performance when dealing with images with low resolution, poor signal-to-noise ratio (SNR) and complicated motion. In this article, we proposed a deep learning-based method for intracellular vesicle tracking. We trained the U-Net for vesicle localization and motion classification on the simulated datasets, which demonstrated high accuracy. We profoundly improved the performance of particle tracking using motion classification, and quantified the dynamic characteristics of intracellular vesicles according to the tracking results with satisfying outcomes. We anticipate that this novel method would have vast applications in analyzing the dynamics in living cell.

To improve the laser beam quality affected by atmospheric turbulence, adaptive optics (AO) is needed. Nowadays, using simple AO systems to achieve low-latency and accurate beam purification is a hot spot in the direction of AO technology. The traditional AO system is greatly simplified by wavefront sensor-less (WFS-less) AO. However, the current WFS-less methods generally have a serious problem that search algorithms is used, which means a lot of iterative calculations and measurements are needed, and they can result in high latency. In order to solve this problem, this article proposes a novel compensation technique based on deep learning and eigenmodes of deformable mirrors (DM), which can obtain the compensation voltages directly from single frame of far-field intensity images. Compared with the existing WFS-less methods, it does not require any iterative operations and has good real-time performance. A convolutional neural network (CNN) model is built in this article, which takes 224*224 far-field intensity images as input and 67-dimensional eigenmode coefficients of DM as output. We completed the closed-loop experiment based on this method, and almost achieved the same result as the closed-loop correction based on the Shack-Hartmann Wavefront Sensor (HSWS).

Near-infrared (NIR) fringe projection is gradually replacing visible fringe projection in face-scanning because NIR light is less harmful to human eyes and has a higher recognition rate in a special environment. However, since NIR is susceptible to interference from various heat sources and light sources, the NIR fringe image captured by the camera is of poor quality and has low contrast. And the captured low-quality fringe image will directly affect the quality of phase acquisition. Traditional phase acquisition methods, such as Fourier transform profilometry and phase-shifting profilometry, are difficult to achieve both high-speed and high-precision phase measurements at the same time. Therefore, this paper proposes a deep learning based phase acquisition method for NIR fringe projection. By using a deep learning model trained by the deep neural network with powerful learning and computing capabilities, phase extraction can be achieved from fewer NIR fringe images. Moreover, our method can retrieve the phase information with high speed and high quality without additional optimization of the original fringe map.

The polarization characteristic of ocean or atmospheric optical sensor is one of the important factors affecting the accuracy of radiation measurement and quantitative inversion. According to polarization ray tracing algorithm, a calculating model of the polarization sensitivity based on the Muller pupil is proposed. A complete coastal zone remote sensor with low polarization sensitivity is designed, optimized and controlled, by the cooperative design of optical configuration compensating and coating. The result shows that the linear polarization sensitivity is less than 2.5% at B1, and below 1.5% for other four bands respectively.

Atmospheric dispersion has a great impact on high-precision astronomical observation. This paper studies the influence of atmospheric dispersion on the imaging system of large aperture astronomical telescope, and the UMAC-based atmospheric dispersion correctors (ADC) control system is designed. According to the data of zenith distance, temperature, humidity and air pressure, the influence of atmospheric dispersion on the imaging system of the telescope is calculated through the theoretical model of atmospheric refraction. The results show that when the zenith distance is greater than 24.83°, the influence of atmospheric dispersion has exceeded 0.3”, that is the requirement of telescope imaging system. So it is necessary to correct the influence of atmospheric dispersion. ADC adopts the structure of two relative rotation glued prisms. The larger the zenith distance is, the higher the requirement of the relative angle control accuracy of the two glued prisms is. According to the results of optical calculations, when the relative rotation accuracy of the two glued prisms is less than 60”, it can meet the imaging quality requirements of the telescope. According to that, the control system of ADC is designed. The rotation control of atmospheric dispersion correction mechanism adopts two-stage reduction mechanism, the first stage is reduction box, the second stage is reduction gear, and the total reduction ratio is 600:1. In order to realize the synchronous control of the two glued prisms, UMAC is used as the main controller, two servo motors are used to drive the rotation of the two glued prisms, and Renishaw encoder is used as the position feedback for position closed-loop control. The peak-to-peak control accuracy of the relative rotation angle of the control system is less than 20”, which meets the control accuracy requirements of the ADC.

In order to meet the requirements of high-precision positioning and adjustment of the secondary mirror of the telescope, the three-dimensional models of two kinds of 3-UPU parallel mechanisms are established. Firstly, the constraints and degrees of freedom(DOF) of the 3-UPU parallel mechanism platform are compared and analyzed by screw method. The analysis results show that the DOF of the mechanism is related to the spatial position relationship of the kinematic pairs. When the type and number of the kinematic pairs in the mechanism are the same, the DOF is not necessarily the same. Therefore, it is necessary to select the mechanism according to the supporting requirements of the secondary mirror and the motion characteristics of the mechanism. It is concluded that the 3-UPU parallel mechanism with non parallel upper and lower Hooke hinge planes has two rotational DOF and one translational DOF, which can meet the motion requirements of the secondary mirror rotating around and moving along the optical axis, and achieve accurate adjustment and positioning.

The effects of temperature on the distribution of space charge and internal electric field in CdZnTe detectors were theoretically simulated by TCAD software. The mechanisms of space charge distribution in CdZnTe detectors at different temperatures and how to manipulate the internal electric field to achieve good detector performance were discussed in this study. The results demonstrated that an uneven internal electric filed distribution was obtained in low temperature. With increasing temperature, the ionized deep levels tended to capture electrons while unionized deep levels to emit electrons. The sub-bandgap light with a wavelength of 850nm and light intensity of 8×10^-8W/cm² were utilized to manipulate space charge distribution by enhancing optical excitation of electrons from the valence band into the ionized deep donor levels. Thus, a flatter internal electric field were achieved, which greatly reduces the probability of carriers being trapped or recombined by defect levels during charge transport processes and then significantly improves the charge collection efficiency of detectors in low temperatures. This simulation results provide a theoretical basis for the application of CdZnTe detectors in astronomical field.

Microwave Kinetic Inductance Detectors (MKIDs), just like a planar resonance cavity resonating at a microwave frequency, are emerging as a kind of high-sensitivity detector suitable for large format arrays at terahertz (THz) wavelengths. There are two types of MKIDs, namely distributed (or antenna-coupled) MKIDs and lumped-element MKIDs (aka LeKIDs). Various superconducting thin films (such as Al, TiN, NbTiN, NbN, and Nb) have been investigated for MKIDs. They do work so long as the detected photon energy exceeds their energy gap, but their response and noise behaviors are yet to be fully understood. Here we report on the design, fabrication, and characterization of distributed and lumped-element MKIDs made of NbTiN superconducting film. Detailed simulation and measurement results will be presented.

Fiber Fabry-Perot filter (FFP-TF) is one of the key components of the fiber Bragg grating (FBG) demodulation system. Its main principle is to realize wavelength scanning with the inverse piezoelectric effect of piezoelectric ceramics (PZT), but the inherent hysteresis and creep characteristics of PZT make the relationship curve between the transmission wavelength of FFP-TF and the control voltage of the PZT unable. Furthermore, in the temperature-varying environment, the relationship between the transmission wavelength and the control voltage keeps drifting. Aiming at the temperature-induced wavelength drift problem of the tunable optical filter, this paper proposed an improved least square support vector machine (LSSVM) model to capture the internal law of the transmission wavelength drift with temperature, and the BAS-PSO algorithm is employed to search penalty factor and nuclear parameters. Experimental results show that after the optimized least squares support vector machine compensates for the tunable filter's sweep fluctuations, the temperature drift error of the tunable filter is ±0.77 pm, and the standard deviation is 0.35 pm, which improves the temperature stability of the tunable filter demodulation in a variable temperature environment.

In order to improve the accuracy and speed of cat-eye target recognition, a cat-eye target recognition method combining fully convolutional residual network and visual saliency is proposed. This method is based on semantic segmentation. Firstly, the collected active and passive images were preprocessed, and saliency detection is used to enhance the cat-eye target in the different image. Second, the active image is segmented by the full convolutional residual network to achieve background suppression and target enhancement, and the fusion result graph containing the cat-eye target region is obtained by the and operation of the two results. Finally, the candidate target area of the cat-eye target Recognition, to extract real cat-eye target. The experimental results show that the proposed method has high accuracy and fast running time, and can be used for real-time detection.

The surface ship target recognition technology based on visual perception is an important research direction in the development of maritime unmanned systems, for the reason that it is the main technical means to ensure the reliable completion of tasks of maritime unmanned systems such as the shipborne unmanned aerial vehicle or the unmanned surface vehicle. In recent years, deep learning technology, especially the deep convolutional neural network, performs well in image classification, target recognition and other tasks, introducing it into the ship target recognition field will promote the breakthrough of ship target recognition technology. Many researchers have introduced the deep convolutional neural networks into the field of ship target recognition and achieved good recognition results. However, due to the fixed position sampling mode of convolution operation and the limitation of the receptive field range in the convolutional neural network, the convolutional neural network generally only extracts the feature information related to the target itself, ignoring the interaction information between different targets and between the target and the scene, thus it has poor adaptability to objects’ spatial geometric transformation, which will affect the recognition performance of ship targets with different scales and different heading directions under occlusion. The human visual perception system can recognize the target quickly and accurately when faced with target scale changes, brightness changes, shape changes, and target occlusion, which largely depends on its inherent visual attention mechanism. Aiming at the problem that the performance of the ship target recognition method based on the convolutional neural network is greatly reduced in the occlusion situation, a convolutional neural network model based on the biological visual attention mechanism was constructed, which can recognize the ship targets with different scales and different heading directions under occlusion quickly and effectively. The model used the residual module with dilated convolution to expand the receptive field of the high-level convolution kernels in the basic feature extraction module and integrate more contextual information into the high-level features. The visual attention module quickly extracted features which were highly related to the target and the current task, thus improving the efficiency and enhancing the model’s adaptability to the geometric transformation of the target space. The multi-scale feature fusion module enhanced the features’ comprehensive expression ability, improved the model’s adaptability to the target scale transformation, and reduced the calculation amount of target location and category prediction. The non-maximum suppression algorithm used the re-scoring mechanism to improve the accuracy of target location and category prediction. The ship target recognition results in the ship target test set with different scales and different heading directions under occlusion which obtained by the proposed method and those of the four mainstream methods based on convolutional neural network were compared, the comparison results show that the average ship target recognition accuracy of the ship target recognition method based on biological visual attention mechanism is improved by 17.51% when comparing with the method which has the highest average recognition accuracy within the four mainstream target recognition methods, its average ship target recognition accuracy reaches to 87.69% which shows strong robustness, the recognition rate meets the real-time requirements meanwhile. The above results show that the proposed method effectively solves the problems of poor adaptability to spatial geometric transformation and loss of valid information caused by the fixed position sampling mode of convolution operation and the limitation of the receptive field range.

In this article, the measurement method of convex hyperboloid using Hindle sphere is further studied. Firstly, the precise analytical formulas for solving radius of curvature, the diameters of outer edge and inner hole of Hindle sphere mirror are developed. The formula can be degenerated to the common approximate formula under certain conditions. In addition, the semi exact analytical formulas for solving the parameters of auxiliary Hindle sphere mirror are derived under some conditions, and the approximate formula is further extended to other occasions of aspheric mirror test, and a feasible stitching detection scheme for the secondary mirror of the CFGT (China Future Giant Telescope) is given. Then, based on the aberration theory, using Hindle auxiliary sphere, the relationships and formulas of the measurement error of vertex radius of curvature of convex hyperboloid mirror caused by the adjustment of interferometer position, the mirrors spacing, and machining error of radius of curvature of the Hindle sphere mirror are obtained, and the Zemax optical software is used for simulation verification.

Microwave kinetic inductance detectors (MKIDs) are promising low-temperature superconducting detectors because of high sensitivity, easy frequency-domain multiplexing (FDM) readout, and simple structure for large-format arrays. Each pixel of the MKID array is a microwave resonator, and the FDM technology makes the resonant frequencies of all resonators read out by one feedline. However, there are often crossovers, missing, and overlapping phenomena between resonance curves. As a prerequisite for imaging, it is necessary to confirm the correspondence between the resonance frequency point and the physical position of the pixel. In this paper, by designing and using an 8×8 LED dot matrix, a fast pixel recognition of an 8×8 MKID array is realized. In addition, the I-V characteristics of the AlGaInP red Light-emitting diode (LED) used in the experiment at different low temperatures are characterized. Finally, through the MKID that has achieved pixel positioning, and the image of LED multi-point lighting is obtained.

A dual-channel focal plane with optical mechanical and electro thermal highly integrated designing is proposed. Firstly a light splitter is applied in the light path, which split the effective light to two focal plane channels with different field of view. To ensure the focusing consistence of these two focal planes, the team splitter is loaded on focusing mechanism, which changed the focusing location of the light, thus the dual-channel focal planes could be focused. The two focal planes have different detectors which give different spectral bands. There are nine detecting spectral bands in all, which is double of the single-channel focal plane camera. With dual channel, multispectral detecting is realized. The spectral range is from visible to near-infrared. This dual-channel focal plane camera is on orbit for more than one year, and it gives fine photos.

Due to the limitation of the existing technology and raw materials, the length of a single CCD and the number of spectral segments that can be integrated on a single CCD are all limited. The optical splicing method of all reflection and all transmission is usually used. In order to increase the number of spectral segments, the focal plane is usually divided into multiple channels by using a spectroscope. To avoid the introduction of the beam splitting components, a splicing and registration method of multispectral focal plane is proposed. First of all, the method and principle of optical splicing and registration are analyzed. Then the form of nine segments focal plane is designed. A single 4-band detector and a single 5-band detector are assembled into a 9-band focal plane according to the field splicing method, and then multiple 9-band focal planes are assembled into a 9-band long linear array focal plane. And the corresponding multi-segment light source is designed in combination with the existing splicing equipment. The light source can not only realize the convenient switching of multiple spectral segments, but also improve the image brightness and detail resolution. Finally, the splicing test of nine segments is carried out and test results are given: the straightness of focal plane splicing is 1.0μm, the coplanar accuracy is 7μm, and the multispectral registration accuracy is 0.5μm, which is equivalent to that of single four and five segment splicing focal plane. The results show that the method is reasonable and feasible, and the accuracy meets the acquirements.

In order to solve the lack of scale information in the camera pose estimated by the monocular catadioptric panoramic camera visual odometry, we propose a binocular catadioptric panoramic camera visual odometry method. By analyzing the projection model of the catadioptric panoramic camera, the calculation method of spherical parallax is analyzed in the binocular panoramic system, and then the depth information of the space point is calculated, and the camera's pose is solved through the 3D-2D feature point matching. The experimental verification on the public data set shows that our method is 23% more accurate than the monocular catadioptric panoramic camera visual odometry method.

The histomorphology of retina is closely related to some common human diseases, such as glaucoma, macular degeneration. The use of deep learning-assisted diagnosis reduces the rate of misdiagnosis and early screening of diseases. There are several difficulties in retinal vessel segmentation as follows. Small vessels located at the end of branches are difficult to be discerned by human eyes. Camera illumination is insufficient or overexposed, resulting in too bright optic disc area, low contrast and blurred retinal vessel side inspection. The unique tree bifurcation structure of retinal vessels is difficult to maintain the original appearance of the structure because the vessels are too thin to be detected. In this paper, we use a U-net network with a stacked full convolutional structure to achieve accurate segmentation of retinal vessels. The main work is as follows. Firstly, the original data are preprocessed: the database images are RGB images, and in order to improve the segmentation accuracy, the channels are extracted for preprocessing first. Secondly, CLAHE is performed to enhance the contrast of the vascular region. Finally, the data is fed into the network for training. U-net is a modified network model of FCN, which mainly consists of feature extraction and upsampling. The feature extraction is used to capture the contextual information in the image, and the upsampling part is used to recover the location information of the image. Compared with the existing algorithms, the proposed algorithm can segment retinal vessels more effectively its sensitivity and accuracy have been significantly improved.

As a mature product with high commercialization, the fiber optic gyroscope is susceptible to the influence of environmental factors in actual use, which affects the measurement accuracy. In order to improve the temperature adaptability of the fiber optic gyroscope and improve the efficiency of temperature compensation in the engineering process of the gyroscope, a temperature modeling and compensation method based on a multilayer perceptron is proposed. First, based on the working principle of the fiber optic gyroscope, the mechanism that causes the temperature error of the fiber optic gyroscope is analyzed. Then, based on the neural network model of the multilayer perceptron, the structure design of the temperature compensation model of the fiber optic gyroscope is carried out, and the existing data is used to train the model. Finally, the compensation model was verified by experiments. The results show that the bias stability of the gyro can be improved by 80% after compensation using this model. Although this method requires a lot of calculations in the early stage, after the model parameters are solidified, it has strong adaptability, is easy to implement in engineering, and can effectively improve engineering efficiency.

Coaxial optical system has a symmetry of revolution. Alignment for this kind of optical system is easy. The desired image quality can be rapidly converged. As for off-axis optical system, traditional optical alignment method can not be used due to the loss of rotational symmetry. Low initial position accuracy makes installation and adjustment more difficult than usual. In this paper, we aim to solve the alignment problem for off-axis optical system with the help of machine learning and its powerful numerical fitting ability. We carried out our research on alignment method for an Gregorian off-axis system. The location of primary mirror is fixed as the optical reference. Alignment process is to adjust posture of secondary mirror to acquire ideal image quality. We use Zemax and Python co-simulation technology to get simulated data. Then multi-layer artificial neural network is utilized to fit the mathematical relationship between misalignments and Zernike coefficients. Given the coefficients, the misalignments can be calculated by the neural network. Finally we conduct alignment experiment to verify the proposed method. The result has proved that this method is a fast and efficient alignment solution for the off-axis optical systems.

With the rise of the new generation of artificial intelligence technology, the object detection method based on deep learning has achieved remarkable results. In this paper, the detection accuracy of three popular object detection algorithms such as You Only Look Once (YOLO V3), Region-CNN (Faster R-CNN) and Single Shot MultiBox Detector (SSD) has been compared. Aiming at the actual detection problems of building block parts with irregular shape and different sizes, a method that combines deep convolutional generative adversarial networks (DCGAN) with deep learning based object detection algorithm is proposed to solve the problems of over fitting or weak generalization ability in the case of limited datasets, and to improve the detection accuracy of object detection algorithm. Experimental results show that: 1. Using public datasets, when the training data is reduced, the mean average precision (mAP) values of the above three algorithms are reduced respectively. Among those, SSD algorithm has the smallest change, which decreases 7.81%. 2. The control variable method is used to train the building block parts. In the case of insufficient training data, the detection accuracy of three object detection algorithms is low. 3. After combining SSD algorithm with DCGAN algorithm and applying it into the detection task of building block parts, the mAP value is improved from 79.63% to 83.32%, and the detection accuracy is obviously improved.

SRCNN firstly applies convolutional neural network to image super-resolution reconstruction, which is the most representative method of super-resolution reconstruction algorithm based on deep learning. In this paper, SRCNN was taken as an example to discuss the application of deep learning method in the super-resolution reconstruction of Wide-Field Infrared and Low Light Level Images. The main work was to compare the effects of training data sets on reconstruction results. Two kinds of data sets were used to train SRCNN. Model 1 used 91 ordinary natural images as training data set, and model 2 used 29 ordinary natural images and 62 Wide-Field images as training data sets. Two groups of Wide-Field Infrared and Low Light Level Images were tested by using the models trained from the two datasets, and the PNSR and SSIM parameters of the test results were compared.

Wavelength tuning laser interferometry can measure the front and rear surface profile and thickness variation of transparent plate at one time. Separating the collected aliasing fringe patterns containing multi-surface interference information can obtain the surface shape information of each surface of the transparent plate. However, in the process of image acquisition and transmission, it will inevitably be affected by noise, and the existence of noise will affect the separation of multi surface shape information, and further affect the recovery of each surface phase and the accurate acquisition of three-dimensional shape. In this paper, a noise reduction method of aliasing fringe pattern based on convolutional neural network is proposed. The simulation data and experimental fringe patterns show that the network can effectively improve the quality of fringe patterns, has faster calculation speed.

Space-based solar observation has severe requirements for resolution, dynamic range, and signal-to-noise ratio of the camera. In order to acquire high-quality solar image data, this paper proposes a high-resolution electronics system based on Gpixel GSENSE6060 image sensor for space-based solar observation. The system uses XILINX XQ5VFX130T as the timing control of the overall system, with DDR SDRAM to cache the image data, which can realize flexible working mode with the windowing mode of the sensor. Firstly, the principle of system parameter selection are given, and the work characteristics of GSENSE6060 are described, then the triggering and termination of event mode are realized by algorithm. The system has high flexibility and reliability, which is suitable for long-time Full-Disk observation and solar eruptions monitoring. During the flare eruption, a high frame rate acquisition with a resolution of 1024 × 1024 can be realized every 4s for the eruption region, which can be used to acquisition the maximum effective data. Experiments show that the system readout noise is better than 6 e^-, in Rolling HDR mode can synthesize 16-bit, resolution of 4608 × 4608 and dynamic range larger than 90dB images, to meet the system design index.

This paper presents an automatic optimization method in ZEMAX sequential mode. The user-defined evaluation function and optimization program are written in macro language, which is used to design the Offner null compensation optical path system. The system is designed with a 1m aperture parabolic lens as an example, of two-piece Offner compensation mirror and the optimal parameters of Offner compensator are obtained through optimization. The compensation residual PV value is better than λ/10 and the RMS value is better thanλ/50. This method is a promising attempt of automatic design of compensator, which can design multiple groups of structures on the basis of improving design efficiency, and make the optimal structure which is more closer to actual demand.

In recent years, artificial intelligence has achieved unprecedented development, and deep learning, represented by neural networks, plays an important role. After the emergence of large-scale pre-trained models with trillions of parameters, the model performance is significantly improved while the burden of computational resources and energy consumption of hardware devices are also increased simultaneously, thus limiting its application in more practical scenarios. Compared with neural networks implemented based on electronic devices, those implemented based on optical devices are called optical neural networks, which have unique properties to overcome the dilemma above. One of the most representative works of optical neural networks these years is the diffractive deep neural network (D²NN). In this paper, the research progress of D²NNs is summarized in four aspects: basic theory, further analysis, improvement, and application. Besides, it is analyzed that the common defect of D²NNs from simulation to physical fabrication, and corresponding theoretical improvement method is also proposed. Meanwhile, to further reduce the impact due to the gap between simulation and physical implementation, and to enhance the robustness of the model, the D²NN training method based on generative adversarial network (GAN) is proposed. The D²NN combines optical transmission with deep learning to achieve complex pattern recognition tasks in the optical domain at the speed of light. It is believed that under continuous research, the D²NN can play a greater role in optical communications and other fields.

In order to improve the measurement accuracy and detection ability of the star sensor, this paper uses a double Gauss structure with conic aspheric surface and hybrid refractive-diffractive to design an optical system with large field of view of 25°and large relative aperture of 1/1.156, and only 6 lenses are used. First, According to the selected CMOS APS sensor performance parameters, the entrance pupil diameter, field of view, focal length and wavelength range are determined. The field of view is 25°, F number is 1.156, focal length is 52mm, entrance pupil diameter is 45mm, the wavelength range is 500-800nm. The detection of 6 magnitude stars is realized, the probability of capturing 5 navigation stars is 100%, and autonomous navigation recognition is realized. Then, the glass material selection and structure optimization of the optical system are conducted, the optical system is optimized by introducing two conic aspheric surfaces and a diffractive surface. Finally, the influence of the diffraction efficiency of the diffractive optical element on the signal-to-noise ratio is analyzed. After analysis of image quality, the shape of diffuse spot in each field of view is approximately circular, the energy is close to Gaussian distribution, more than 90% of the energy is concentrated in 3×3 pixels, lateral color is less than 1.06μm, and relative distortion is less than 0.87%. The optical system only uses six lenses, the imaging quality meets the imaging requirements of the star sensor, and is beneficial to improve the accuracy and detection capability of the star sensor.

In fringe projection profilometry (FPP), efficiently recovering the absolute phase has always been a great chal-lenge. The stereo phase unwrapping (SPU) technologies based on geometric constraints can eliminate phase ambiguity without projecting any additional patterns, which maximizes the efficiency of the retrieval of abso-lute phase. Inspired by recent successes of deep learning for phase analysis, we demonstrate that deep learning can be an effective tool that organically unifies phase retrieval, geometric constraints, and phase unwrapping into a comprehensive framework. Driven by extensive training dataset, the properly trained neural network can achieve high-quality phase retrieval and robust phase ambiguity removal from only single-frame projection. Experimental results demonstrate that compared with conventional SPU, our deep-learning-based approach can more efficiently and robustly unwrap the phase of dense fringe images in a larger measurement volume with fewer camera views.

In order to improve the multi station measurement accuracy of laser tracker, aiming at the problem that the measurement error of laser tracker restricts the transfer accuracy, the compensation method of measurement error of single laser tracker in multi-station measurement is studied in this paper. The compensated coordinate measurement values are substituted into the coordinate transformation model, and the weighted total least square method is used to iteratively solve the transfer parameters. The results are verified by Matlab simulation experiments. The results show that the mean square errors of coordinate points before and after compensation are respectively 0.4358μm and 0.2920μm. The data show that the measurement accuracy of the laser tracker is improved by 30%.

Astrophotonics aims to apply photonic technology to manipulate light from a telescope and process it for astronomical research purposes in an efficient and cost-effective way. In particular, using planar dispersive device for astronomical spectroscopic research is a promising method, as the compact and lightweight device could replace at least some of the bulky parts in the modern spectrograph. However, the migration of arrayed waveguide gratings from optical communication to astronomy must go through new designs with tailored properties. In this work, we first demonstrate an arrayed waveguide grating with over 100-nm free spectral range and further propose a new design of an integrated photonic spectrograph without any free-space optics. The design merges the wavefront modulation of a conventional lens into the length/phase variation in the waveguide array, in addition to the required difference to form the grating effect. The end effect is that the output light from the facets of the arrayed waveguides is not only separated by wavelength, but also focused onto a camera at a fixed distance. Simulation shows this design can reach high resolving power (> 7,000) with more compact size and most importantly, it eliminates the need to add any conventional lens in the optical path, in the sense of a truly integrated photonic spectrograph.

Diffraction grating is an important optical device whose surface defects will seriously affect the quality of optical system.In order to achieve the diffraction grating of the ultra precision surface defect inspection, design a set of defects automatic detection system based on machine vision, we use black hat transform images highlight several images, and after a Canny edge detection, expansion, corrosion, determine the connected domain, look for the seed point, the algorithm of region growing process, realize the defect feature extraction, classification and statistics.A two-dimension displacement platform controlled by DSP is designed to realize the panoramic image Mosaic of diffraction grating and accurately locate the coordinates of defects.At the same time, the surface quality of the grating is evaluated according to the American military optical appearance standard MIL-PRF-13830B.

As an essential component of the slitless spectrometer, the UV transmission blazed grating has the capability of high dispersion and high resolution. In this paper, a method for fabricating UV transmission blazed gratings by holography-ion beam etching is proposed. Holographic interference lithography is used to generate photoresist grating masks. The ion beam vertical etching transfers the photoresist mask pattern to the substrate to form a SiO₂ grating mask. When the ion beam incident direction is at a certain angle to the normal direction of the substrate, the SiO₂ mask is used to block the inclined ion beam, so that different parts of the mask bottom are bombarded by the ion beam with different fluxes, forming a blazing facet. When the mask is etched completely, the blazed grating is formed. Based on the idea of the line motion algorithm, the article establishes the geometric model of blazed grating etching, which provides the parameter guidance for precise control of the groove structure. Combined with the theoretical model, a UV transmission blazed grating with a line density of 333 lines/mm and a blazing angle of 13.2° is successfully fabricated.

In this article, we designed and fabricated a high linear density concave variable line space holographic grating for the Lyman spectrometer, with level 3 working and the center line density is 3300line/mm. The focus curve is a circle with the grating vertex O as the center and a radius of 900 mm. We analyze the influence of the exposure error on the line density distribution and reduce the harm of errors through the method of multi-error compensation. To improve the diffraction efficiency, we use Finite Element Method software to get the best grating groove parameters. Finally, we initially prepared a holographic grating with a symmetrical arch groove with a groove depth of 175nm and a bottom duty cycle of 0.3.

We design and fabricate a 350GHz 8×8 Al Microwave Kinetic Inductance Detector (MKIDs) array for the demonstration of its characteristics, mainly focus on the quasiparticle lifetime of the resonators. The quasiparticle lifetime data is collected by measuring a resonator’s phase response to a LED pulse at resonant frequency in a dilution refrigerator cooled to mK stage. We also measure and discuss the influence of various parameters on the change of quasiparticle lifetime, including different LED voltage supply, bath temperature of the MKIDs, and superconducting film thickness.

Phase retrieval is of great significance in the fields of medical imaging and computational holography. To solve the problem more efficiently, this paper presents a phase recovery network with two modes of operation based on the convolution neural network, which not only can get persistent model by training data set, but also can build a special loss function to recover the unknown signal in a self-optimized way without data set in the case of Gaussian measurement model. Comparison of the simulation results show that the network is able to obtain better results with fewer measurements than the existing phase recovery algorithms.

Off-axis refractive system with the noticeable advantages such as high resolution, large view field and central obscuration removed, has been one of the powerful systems for space astronomical telescopes in recent years. However, misalignment errors and surface error of mirrors are significant especially in the alignment progress of off-axis reflective telescope with large aperture. Computer aided assembling (CAA) jointly provide a robust misalignment correction method to ensure the accurate alignment of telescope. In this paper, system aberration of misalignment coaxial system with two mirrors is analyzed in detail, moreover, the off-axis system is studied further, especially in the off-axis Gregorian system. And the feasibility of correction values solution about off-axis refractive system is discussed. Both the simulation and experiment results demonstrate the feasibility of the proposed alignment method and high accuracy has been achieved. In the testing off-axis Gregorian system, the primary mirror is paraboloid with 1200 mm diameter, 210 mm off-axis distance, and the second mirror is ellipsoid with off-axis distance 129.0 mm, focal length 425 mm and 2125 mm, respectively. For the testing off-axis Gregorian system, the RMS value of primary mirror and second mirror are 0.021 λ and 0.027 λ (λ = 0.6328 nm), and the testing optimization result of system wavefront aberration with RMS value is better than 0.058 λ is achieved. The reverse optimization method testing can achieve high-accuracy measurement ability, which provides efficient and flexible way for the off axis refractive system from various types of elements with complex surfaces.

The temperature drift causes the zero-bias drift of the fiber optic gyroscope to show complex nonlinear changes, which seriously restricts the measurement accuracy of the fiber optic gyroscope. Therefore, it is necessary to establish an accurate temperature compensation model to compensate for the temperature drift of the fiber optic gyroscope.In order to effectively improve the output accuracy of the fiber optic gyroscope under the condition of the full temperature range, the static full temperature bias test of the fiber optic gyroscope is first designed to obtain the bias data of the fiber optic gyroscope under the temperature change condition of -40℃~60℃. Secondly, on the one hand, a polynomial regression model is gradually established with temperature, temperature change and multiple powers as independent variables. On the other hand, the RBF neural network model is established after screening the input variables with the MIV algorithm. Finally, two models are used to achieve zero-bias temperature compensation. According to the compensation results, both can effectively improve the full temperature output accuracy of the fiber optic gyroscope. Compared with the polynomial regression model, the RBF neural network model can identify temperature drift more effectively and accurately, and greatly improve the output accuracy of the fiber optic gyroscope in the full temperature range.

To achieve precise applications under specific environmental requirements, an infrared optical system that can be applied to near-infrared wavelengths was designed. First of all,based on the selected parameters，the initial structure was selected and the form of the Maxutov telescope was selected and improved.Then，the detection performance of the design results was analyzed.Due to the large range of high and low temperature changes in the outer space environment, the system lens barrel material analysis was carried out to obtain the influence of temperature changes on the energy distribution of the optical system.Finally，a tolerance analysis and optimization of the designed system were conducted so that it could meet the needs of processing and assembly.The design and analysis results indicate that the optical system adopts a spherical catadioptric mirror surface, the total length of the system is 126.218mm, the full field of view for a 14 μm surrounding energy distribution is > 80%, the maximum magnification chromatic aberration value of monochromatic light with a center wavelength of 1.3μm is < 4μm, the maximum absolute distortion is < 4μm in 0.8 field of view.The system has a large field of view，aperture，and relative aperture. It is compact and of suitable size，easy to install，and offers high detection performance，high detection sensitivity，and a wide detection range.The optical system is useful for the accurate detection of high-precision target detection in a wide spectrum covering the near-infrared band of 0.9~1.7μm.

In this paper, the deployment systems of a deployable space telescope for CubeSat are proposed. This telescope can achieve a ground resolution less than 1 meter at an orbital altitude of 400 km. Both the primary and secondary mirrors of it can be stowed in 3U volume（100mm×100mm×300mm）during launching period. The functions of the deployment systems include the locking of the primary and secondary mirrors when stowed, the deployment on-orbit, and the active adjustment of the primary mirror segments after deployment. In this paper, firstly, the optical system of deployable telescope is described, then the deployment systems of the telescope are designed. Finally, the modal analysis is carried out using finite element method.

In recent years, astronomical observation has put forward higher requirements for observation environment, resolution power and simultaneous observation quantity, which makes the cost of traditional spectrometer spiraling. The development of integrated photonics provides a miniaturized, low-cost and environment-friendly solution for astronomical spectral observation. However, high resolution spectrum is hard to achieve because of the highly repetitive structure and defocus aberration existing in re-imaging system when needing cross dispersion. Based on Silica-on-Silicon material platform with a refractive index contrast of 0.23%, new structure of cascaded modulation arrayed waveguides(CMAW) with total length difference of 22mm are arranged in a square chip with a side length of 40mm, having a theoretical resolution of 20,000@1550nm and tested resolution of 15,000@1550nm. Our work demonstrates the potential of integrated photonics applying in high resolution astronomical spectral observation, three times above the record of available resolution power international, which is expected to provide a solution for space observation and large-scale spectral survey.

As a hot research direction in the field of communication reconnaissance, signal modulation classification plays an increasingly important role in the field of national defense. Traditional signal modulation style classification methods are mostly based on the combination of feature engineering and pattern recognition. First, the expert features are extracted by manual design, and then the signal modulation is recognized by the pattern recognition algorithm. The limitation of this method is that an expert feature can only effectively identify a few specific modulation signals. Besides, the deficiency of the number of expert features will lead to a low classification accuracy. In order to improve the accuracy of signal modulation classification, a signal modulation classification algorithm based on convolutional neural network is proposed. Convolutional neural networks can achieve end-to-end classification without manually designing and extracting features. Convolutional neural networks can automatically extract various levels of abundant features through learning, which can improve classification accuracy. The convolutional neural network architecture designed in this paper includes: three convolutional layers, three pooling layers, and the last layer is the softmax classification layer, which outputs the classification results. Experimental results show that on a data set containing 32 I/O signals, when the signal-to-noise ratio is 6dB, the algorithm has a training accuracy rate of 92.7% and a test accuracy rate of 90.2%.

The difficulty or even the impossibility of manufacturing of lenses with large apertures results in optical systems holding at least two mirrors, serving as the primary and the secondary mirrors respectively, are widely used in space application to realize large apertures, long focal lengths and high resolution, etc. Some of these optical systems fall into the catadioptric group, and others are all-reflective ones. What is more, the latter ones take the advantage of chromatic-aberration-free and the possibility of further lightening. To get an achromatic, light-weighted and compact zoom optical system with a large aperture, a long focal length and high solution, the all-reflective zoom system becomes an ideal choice. A method is described for the design of mechanically compensated reflective zoom systems with 3 mirrors. The method is based on the 3rd-order aberration theory. The advantage of this method is that, through constrained optimization of a set of Seidel aberration coefficient functions, it allows the designer to achieve the initial construction parameters of the optical system, and the design results prove the feasibility of this method.

To realize highly integrated and portable spectrometer, a new type of computational spectrometer scheme has been proposed in recent years. In the existing computational spectrometers, many algorithms have been used for the inversion of the input spectra, such as principal component analysis (PCA), Tikhonov regularization inversion, compressive sensing and so on. However, A huge amount of data will probably result in a calculation failure. In order to solve the problem, this paper proposed a random pixelated grating computational spectrometer based on deep learning. The computational spectrometer uses random pixelated grating as scattering medium and records the wavelength-dependent speckle pattern. Meanwhile，the fiber spectrometer is used to record the wavelength of the incident light. Meanwhile，6750 speckle patterns corresponding to various wavelengths are recorded in the calibration experiment. we use the PyTorch framework to build a convolutional neural network architecture of deep-learning ResNet50. We use 6600 speckle patterns as the training set. After the model is well trained by the speckle patterns, it is then used to predict the wavelengths of 150 speckle patterns. The experimental results show that the algorithm can successfully reconstruct the output spectrum of the laser diode (LD) with single longitudinal mode. the system resolution is about 0.4nm, and the correctness of wavelength prediction can be up to 95%. The experimental results show that the deep learning algorithm can be used for wavelength inversion of the computational spectrometer, and it has the advantages of high inversion speed and large data processing ability.

Optical imaging satellite constellation (OISC) can provide imagery products with optimized coverage performance and high temporal resolution than single satellite, and is widely used in target detecting and analysis. People highly concerns with the performance in target identification of OISC during observation missions, so it is necessary to evaluate the target characteristics (TC) detecting capability of satellite constellation. Based on Analytic Hierarchy Process (AHP), a detection capability evaluation method for OISC is presented in terms of TC in this paper. A hierarchical structure is established to demonstrate the affiliation of indexes concerned with TC and capability of OISC. After that, weight coefficient of each index is computed using AHP. Moreover, we put forward quantitative method to normalize performance indexes of OISC, and calculate the evaluation result by linear-weighted summation method. At last, Jilin-1 Constellation, Zhuhai-1 Constellation, etc., are evaluated with the method to test its feasibility.

Generative adversarial network (GAN) has become a hot research topic in the field of image processing. As an unsupervised training model, GAN has been widely used in the field of computer vision, especially in image style transfer. The purpose of the GAN is to make the generator generate a false image, and the discriminator cannot tell whether the input image is the real image or the generated image. Compared with traditional network models, GAN model has these advantages in image style transfer: GAN is composed of two different networks, and the loss function is automatically learned by playing games with each other. GAN belongs to unsupervised training and does not need to annotate the data set, which saves a lot of work. In this paper, improved GAN models related to image style migration are summarized. Firstly, the principle and method of image style transfer based on convolutional neural network are introduced. Secondly, the status, principle and prospect of GAN are introduced, and the causes of gradient disappearance and mode collapse of GAN are analyzed in detail. On this basis, the principles, advantages and disadvantages of CGAN, DCGAN, CycleGAN and StarGAN V2 network models are introduced. Finally, it summarizes the current problems and future research directions of style transfer based on GAN.

The fiber-fed High Resolution Spectrograph of the Chinese Xinglong 2.16-m Telescope is the main instrument for high precision spectral detection in China. In order to improve the observation performance, a wavelength calibrator based on Fabry-P´erot etalon (FPE) is developed. Here, we report on the design, transmission charac-teristics and vacuum-thermostatic performance of the system. To meet the requirements of the calibration, the calibrator is custom-designed with the FPE gap of 5 mm, fineness value of 40, and illuminating fiber of 100 μm in diameter, which could generate calibration lines with 30 GHz in lines space and covering 500 – 700 nm.

Integrated optical phase-shifting interferometer is one of key devices that are frequently employed in astronomical interference instruments to measure complex coherent function of celestial object at certain spatial frequency, and further to achieve high-resolution imaging by synthetic aperture technology in last decades. Due to the presence of device fabrication error, calibration of an interferometer is indispensable, Pixel-to-Visibility Matrix (P2VM) method being a conventional approach. Up to present days, the P2VM is employed as a “all-in-one” method, which unify all device errors into one single matrix. However, characterization of device fabrication error at its various parts is often desired, to provide the basis for device optimization. In the current paper, a process to retrieve fabrication errors of the three main parts of the phase-shift interferometer, from an experimentally measured Pixel-to-Visibility Matrix (V2PM) is demonstrated. The results are expected to be useful in the process of optimizing the device structure and its manufacturing parameters.

The difficulty or even the impossibility of manufacturing of lenses with large apertures results in optical systems holding at least two mirrors, serving as the primary and the secondary mirrors respectively, are widely used in space application to realize large apertures, long focal lengths and high resolution, etc. Some of these optical systems fall into the catadioptric group, and others are all-reflective ones. What is more, the latter ones take the advantage of chromatic-aberration-free and the possibility of further lightening. To get an achromatic, light-weighted and compact zoom optical system with a large aperture, a long focal length and high solution, the all-reflective zoom system becomes an ideal choice. A method is described for the design of mechanically compensated reflective zoom systems with 3 mirrors. The method is based on the 3rd-order aberration theory. The advantage of this method is that, through constrained optimization of a set of Seidel aberration coefficient functions, it allows the designer to achieve the initial construction parameters of the optical system, and the design results prove the feasibility of this method.