2023
2023
-
Record 109 of
Title:Research on Infrared Nonuniformity Correction Method Based on Transformer
Author(s):Ji, Ran(1,2); Xiao, Maosen(3); Liu, Yu(1,4); Cheng, Jiawei(1,4)Source: 2023 International Conference on Image Processing, Computer Vision and Machine Learning, ICICML 2023 Volume: null Issue: null Article Number: null DOI: 10.1109/ICICML60161.2023.10424943 Published: 2023Abstract:Aiming at the problems of low generalization ability and high system complexity in the existing infrared image nonuniformity correction methods, this paper proposes a single-frame infrared image nonuniformity correction method based on Transformer, which has the advantages of simple network structure, strong model generalization ability, etc., and the average residual nonuniformity of the images in the test set can reach below 5%. ? 2023 IEEE.Accession Number: 20241015682575 -
Record 110 of
Title:Quasinormal Mode Expansion Method for Resonators with Partial-fraction Material Dispersion
Author(s):Ming, Xianshun(1)Source: arXiv Volume: null Issue: null Article Number: null DOI: 10.48550/arXiv.2312.11048 Published: December 18, 2023Abstract:In this paper, we first establish a Quasinormal Mode (QNM) solver for open resonators made of materials with general dispersion which can be modeled by partial fractions, and develop the corresponding analytical QNM expansion method (QNMEM) for both discrete and periodic resonant structures. When the response of the resonators is dominant by several leading QNMs, a simplified QNMEM can be used to analyze their spectra in a reasonable accuracy. The simplified QNMEM is used to analyze the spectra of the metal-dielectric-metal perfect absorber, which has the advantages of both high computation speed and clear physical insight. ? 2023, CC BY.Accession Number: 20230459152 -
Record 111 of
Title:Deep image prior for polarization image demosaicking
Author(s):Shi, Yinxia(1); Wen, Desheng(2); Jiang, Tuochi(2)Source: 2023 4th International Conference on Big Data and Artificial Intelligence and Software Engineering, ICBASE 2023 Volume: null Issue: null Article Number: null DOI: 10.1109/ICBASE59196.2023.10303066 Published: 2023Abstract:The Division of Focal Plane (DoFP) image sensors can obtain the polarization information of the object surface. However, the polarization images generated by the DoFP image sensors have the problem of instantaneous field of view (IFoV) error and resolution reduction. To solve this problem, we propose a method based on deep image prior to solve this problem. Our method employs direct adaptation of the generator network to a single low-resolution polarized image and does not require training on a large dataset. Comparing our proposed method with other polarization image interpolation methods, our method performs better in visual, peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM). ? 2023 IEEE.Accession Number: 20234915151195 -
Record 112 of
Title:Research on calibration method for mirror attitude and tilt monitoring system
Author(s):Cao, Mingqiang(1); Fu, Xing(1); Xu, Zhichen(1); Ning, Xuanqi(1)Source: Proceedings of SPIE - The International Society for Optical Engineering Volume: 12976 Issue: null Article Number: 1297624 DOI: 10.1117/12.3009616 Published: 2023Abstract:This paper investigates a method for measuring and calibrating the mirror attitude, tilt monitoring system, and compensation lens in a diffraction camera. Firstly, a measurement network is established by installing a reference prism and a theodolite on the load-bearing plate, enabling the overall measurement of the position of the mirror relative to the camera. Then, by changing the attitude of the mirror and recording the changes in the detector point coordinates, the relationship between the direction angle of the mirror and the detector point coordinates is calculated. Then, by adjusting the compensation lens in its local coordinate system, the relationship between the compensation lens and the detector point coordinates is obtained. Finally, by substituting the above matrix into the conversion matrix formula, the conversion matrix between the direction angle of the secondary mirror and the compensation lens translation is obtained, thereby completing the calibration of the attitude. This method can directly drive the compensation lens according to the change in the position of the mirror, ensuring the imaging quality of the camera. ? COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.Accession Number: 20240315401973 -
Record 113 of
Title:High-precision pose measurement method based on binocular vision in dark lighting environments
Author(s):Wang, Feng(1); Zhang, Haifeng(1); Zhang, Gaopeng(1); Shan, Fuqiang(1,2); Ren, Long(1); Ai, Han(1,2); Cao, Jianzhong(1)Source: Optical Engineering Volume: 62 Issue: 2 Article Number: 024105 DOI: 10.1117/1.OE.62.2.024105 Published: February 1, 2023Abstract:Measuring the pose of non-cooperative targets in space is a critical supporting technology for cleaning up space debris and recovering items. However, most existing methods are simulation experiments conducted in good lighting environments and tend to show poor performance in dark lighting environments. A target pose measurement method based on binocular vision is proposed, which is suitable for dark lighting environments. First, the traditional features from accelerated segment test algorithm are improved to reduce the influence of illumination on the performance of feature point extraction under various postures. The point feature and line feature are combined to extract image features more easily in a dark lighting environment while retaining the high accuracy of the pose measurement algorithm based on point features. Second, the normalized cross-correlation coefficient matching method is combined with the epipolar constraint to narrow the search range of the matching points from the two-dimensional plane to the epipolar line, which substantially improves the matching efficiency and accuracy of the matching algorithm. Finally, post-processing through feature matching is performed to reduce the probability of mismatches. Simulation and physical experiment results show that our method can stably extract features and obtain high-precision target pose information in well-illuminated as well as dark lighting environments, making it suitable for high-precision target pose measurement under insufficient illumination. ? 2023 Society of Photo-Optical Instrumentation Engineers (SPIE).Accession Number: 20231013686176 -
Record 114 of
Title:Research on precision alignment method of a five-mirror optical derotator system
Author(s):Lei, Yu(1,2); Xu, Songbo(1); Cao, Mingqiang(1); Ma, Caiwen(1,2); Duan, Zhanjun(1); Fu, Xing(1); Li, Hua(1); Kang, Shifa(1)Source: Journal of Astronomical Telescopes, Instruments, and Systems Volume: 9 Issue: 1 Article Number: null DOI: 10.1117/1.JATIS.9.1.014004 Published: January 1, 2023Abstract:A five-mirror optical derotator system is used in the Accurate Infrared Magnetic System solar telescope by virtue of its polarization-free and superior real-Time performance. The derotator system can compensate image rotation during tracking observation. The system consists of five flat mirrors with their normal vectors noncoplanar. Due to the complicated spatial positions of mirrors, it is challenging to align the system with high accuracy. We analyze parallelism and concentricity characteristic of derotator system by matrix transformation and propose a compensation alignment method from multivariables perturbation simulation. This method reduces degrees of freedom for alignment from 10 to 4, which greatly simplifies the installation and adjustment process. Based on the above simulation, the alignment experiment has been conducted successfully with the parallelism and concentricity meeting the requirements. Through theoretical analysis and experimental verification, the proposed method is reasonable and provides an efficient alignment solution for this kind of five-mirror optical derotator system. ? 2023 Society of Photo-Optical Instrumentation Engineers (SPIE).Accession Number: 20231513865966 -
Record 115 of
Title:Research on assembly and rectification method of complex offset axis infrared diffraction camera
Author(s):Cao, Ming Qiang(1); Liu, Jun Peng(1); Xu, Hong Lao(1); Kang, Shi Fa(1); Fu, Xing(1); Qin, Xing(1); Shen, Zhong(1)Source: Proceedings of SPIE - The International Society for Optical Engineering Volume: 12507 Issue: null Article Number: 125072Q DOI: 10.1117/12.2656677 Published: 2023Abstract:In this paper, the assembly and rectification method of the new optical system is studied. According to the characteristics and difficulties of the diffraction camera, the assembly and rectification scheme is formulated. Firstly, the space angle of the mirror is determined by the spatial attitude measurement and the reference conversion matrix method. The coaxial adjustment and combined detection and adjustment of each transmission lens group are completed by the self alignment method. The method of sensitivity matrix iterative fine adjustment based on aberration is applied in the whole machine assembly and adjustment process. Finally, the high-precision assembly and detection of the complex offset axis infrared diffraction camera were completed, and the wave aberration on the axis reached 0.119λ@3.39μm. The MTF reached 0.13@33lp/mm, meeting the design requirements. This method provides engineering experience and reference for similar camera assembly and adjustment process methods. ? 2023 SPIE.Accession Number: 20230613537988 -
Record 116 of
Title:Target Localization Technology Based on Biomimetic Curved Compound Eye Camera
Author(s):Zhu, Shuaimin(1); Guo, Wenge(1); Liu, Tao(1); Zhang, Yuanjie(2,3); Xu, Huangrong(3); Wu, Dengshan(2); Zhou, Xiaojun(2); Yu, Weixing(2,3)Source: Guangzi Xuebao/Acta Photonica Sinica Volume: 52 Issue: 9 Article Number: 0911003 DOI: 10.3788/gzxb20235209.0911003 Published: September 2023Abstract:Using a vision system to locate a target is a necessary step for its three-dimensional detection of the target. The traditional single-aperture imaging system can only obtain the geometric image information of the target. A compound eye vision system has the advantages of large field of view， large depth of field， multi-channel imaging， and can obtain the depth information of the target and be sensitive to fast moving targets. At present， a common visual positioning method is to use the binocular vision system to locate the target based on the parallax between two cameras. However， because the binocular vision system has only one set of constraints， and the baseline is fixed， the binocular vision system has low positioning accuracy in the long distance， while the compound vision system has more constraints because of the number of sub-eyes. In the long distance， the positioning accuracy is higher than the binocular vision system. It has aroused a wide attention of researchers. This paper uses the bionic curved compound eye camera developed in the laboratory to carry out the research of 3D positioning and 3D reconstruction. The compound eye vision system consists of a curved compound eye， an optical relay image conversion subsystem and a high-definition image sensor. In this paper， CAL Tag calibration board and MATLAB stereo calibration toolbox is used to calibrate the internal parameter matrix of the compound eye camera and the rotation matrix and translation vector between the sub-eye and the world coordinate system. Based on the principle of binocular vision positioning， a mathematical model for multi eye positioning is established on a compound eye vision system developed in the laboratory， and positioning experiments are conducted. The experimental system includes a laser rangefinder， black cardboard， and a compound eye vision system. The laser spot is used as a positioning target. Because the shape of the sub-eye is circular， the hough circle transformation algorithm is used to detect the sub-eye of the compound eye system， and the sub-eye number is determined according to the center coordinates and radius of the circle. Because this experiment is carried out under dark conditions， the background gray value is low and the spot gray value is high， so the gray centroid method is used to locate the centroid of the spot and obtain the centroid of the spot taken by different sub-eyes. The three-dimensional coordinates of the centroid of the spot are obtained from the coordinates of the centroid of the spot in the camera pixel coordinate system according to the corresponding relationship between the pixel coordinate system and the world coordinate system. The linear equations of several sub-eyes are combined to form the overdetermined equations and the optimal solution is obtained by the least square method. The distance measurement experiment results show that the distance measurement error of the compound eye camera is less than 2% within a range of at least 4 meters. The experimental results show that the bionic curved compound eye camera prepared in the laboratory could carry out more accurate three-dimensional positioning of objects in space. The error caused by the laser jitter and the size change of the light spot with the distance change on the positioning result is analyzed in detail. In the aspect of target 3D reconstruction， the sift algorithm is used to detect and match the feature points of the target images of different sub-eyes， and the RANSAC algorithm is used to remove the wrong matching points， to obtain the accurate feature point matching of the target captured by different sub-eyes. Then， according to the corresponding relationship between the pixel coordinate system obtained by camera calibration and the world coordinate system， the three-dimensional coordinates of the feature point in the world coordinate system are calculated from the coordinates of the sub-eye pixel coordinate system， and the complete reconstructed point cloud of the target is obtained through point cloud stitching. The 3D reconstruction experiment is carried out by the reconstruction algorithm. The experiment takes the cube covered with speckles as the reconstruction target. The cube is photographed at about 0.6 meters from the camera， and a relatively complete 3D reconstruction point cloud is obtained. The research results in this paper show that the bionic curved compound eye camera has great development potential and application prospects in the fields of 3D positioning， 3D reconstruction and optical navigation. ? 2023 Chinese Optical Society. All rights reserved.Accession Number: 20234014824836 -
Record 117 of
Title:Research on thermal stability of monolithic near-infrared Doppler asymmetric spatial heterodyne interferometer
Author(s):Chang, Chenguang(1,2); Fu, Di(1,2); Zhao, Hengxiang(1); Hao, Xiongbo(1); Li, Juan(1); Wang, Pengchong(1); Sun, Jian(1); Feng, Xiangpeng(1); Kong, Liang(1,2); Feng, Yutao(1)Source: Optical Engineering Volume: 62 Issue: 1 Article Number: 015104 DOI: 10.1117/1.OE.62.1.015104 Published: January 1, 2023Abstract:A Doppler asymmetric spatial heterodyne (DASH) interferometer was designed to measure atmospheric winds at a height of 60 to 80 km by observing the airglow emission line of molecular oxygen at 867 nm. The designed monolithic DASH interferometer exhibited decent thermal stability. The phase thermal drift of the fabricated interferometer obtained from thermal performance measurements was 0.376 rad/°C. To accurately model and minimize the thermal drift performance of an interferometer in the design phase, it is necessary to include the influence of thermal distortion of the monolithic interferometer components. Therefore, an optical-structural-thermal integrated analysis method based on Zernike polynomials was proposed to accurately calculate the phase thermal drift of the interferometer. The optical model modified by the finite-element method calculated the phase thermal drift to be 0.420 rad/°C, which agreed with the experimental result within 11.7%. This analysis method can accurately calculate and optimize thermal stability during the design of a DASH interferometer. ? 2022 Society of Photo-Optical Instrumentation Engineers (SPIE).Accession Number: 20230613566289 -
Record 118 of
Title:Infrared small target detection algorithm based on entropy weighted multiscale local contrast
Author(s):Wei, Jingbo(1,2); Chen, Rongli(1); Zhang, Ximing(1); Zhao, Hui(1)Source: Proceedings of SPIE - The International Society for Optical Engineering Volume: 12557 Issue: null Article Number: 125570U DOI: 10.1117/12.2648172 Published: 2023Abstract:It is an important and challenging topic to deal with infrared small target detection with high detection rate, low false alarm rate and low computational complexity in various application fields. Aiming at solving the problem that the existing algorithms can not effectively enhance the real foreground target and suppress various complex background interference. This paper proposes an infrared small target detection algorithm based on entropy weighted multiscale local contrast. Firstly, the local contrast formula is redefined in the joint form of ratio and difference, which can enhance the target and suppress the background clutter. Secondly, the local entropy can be used to reflect the gray mutation in the local area of the image. We use the modified local entropy operator to weight the multiscale local contrast. Finally, we employ the adaptive threshold segmentation to separate the target from the background and obtain the final infrared small target. We test six groups of infrared image sequences with different targets and backgrounds, the backgrounds include mountain, forest, field, sea-sky, sky and thick cloud. Experimental results show that, the proposed algorithm can not only robustly detect infrared dim and small targets of different sizes in various complex backgrounds, but also has higher detection efficiency and lower false alarm rate compared with other traditional baseline methods. ? 2023 SPIE.Accession Number: 20230813600591 -
Record 119 of
Title:Research on data processing method for detection of small and weak targets in space
Author(s):Sen, Tian(1,2); Yan, Wen(1,2)Source: Proceedings of SPIE - The International Society for Optical Engineering Volume: 12557 Issue: null Article Number: 1255729 DOI: 10.1117/12.2652025 Published: 2023Abstract:Space debris identification and localization has been studied for a long time at home and abroad, but there are still shortcomings in the detection of very low SNR small targets, especially the detection and recognition of very low SNR faint small targets submerged in the dense galactic background Stars in order to effectively target detection, background removal is essential, in order to distinguish between the target and the stars, through the accumulation of judgement block number of star closely adjacent to judge, if the judgment of the block, star closely adjacent number more than set threshold, so that the point for stars, on the other hand, argue that point as the goal to be detected, then, to the star point of the block The results of the algorithm show that, according to different star map images, the size of matrix block and the threshold of star point can be modified to achieve better star removal effect. ? 2023 SPIE.Accession Number: 20230813600570 -
Record 120 of
Title:Nonlinear image translation using adaptive rectifier with structure adaption
Author(s):Zhang, Yipeng(1,2,3); Hu, Bingliang(1,2); Ning, Hailong(4,5,6); Wang, Quan(1,2)Source: Journal of Electronic Imaging Volume: 32 Issue: 2 Article Number: 023007 DOI: 10.1117/1.JEI.32.2.023007 Published: March 1, 2023Abstract:When an image translation task contains intradomain translations, the untranslated source image will be discriminated as the real by the discriminator. Thus if the network's nonlinearity is insufficient, the generator can fool the discriminator by producing output that resembles the source image. We propose an activation function termed "adaptive rectified linear unit (ReLU) with structure adaption (SA-AdaReLU)"to enhance the control and nonlinearity of the network in image translation tasks. SA-AdaReLU is composed of two technologies: adaptive ReLU (AdaReLU) and structural adaptive function. The proposed AdaReLU can dynamically change the channel-wise data distribution to better utilize the features in negative regions, which helps to improve the control of the network when inner-domain translation is involved. Meanwhile, the structural adaptive function further enhances the feature selection ability of adaptive instance normalization (AdaIN) and enhances the network's spatial nonlinearity to manipulate the spatial structure on the feature maps. Extensive experiments have demonstrated the effectiveness of the proposed SA-AdaReLU. In addition, with SA-AdaReLU, fewer layers are required to achieve the same visual effect for building the generator, thus reducing the computational complexity. ? 2023 SPIE and IS&T.Accession Number: 20232014101757