2020

• Record 73 of
  
  Title:Spectral–temporal hybrid modulation for channeled spectropolarimetry
  
  Author(s):Li, Qiwei(1,2,3); Alenin, Andrey S.(2); Scott Tyo, J.(2)
  Source: Applied Optics  Volume: 59  Issue: 30  DOI: 10.1364/AO.404623  Published: October 20, 2020  
  Abstract:Channeled spectropolarimeters (CSPs) are capable of estimating spectrally resolved Stokes parameters from a single modulated spectrum. However, channel crosstalk and subsequent spectral resolution loss reduce the reconstruction accuracy and limit the systems’ scope of application. In this paper, we propose a spectral–temporal modulation strategy with the aim of extending channel bandwidth and improving reconstruction accuracy by leveraging the hybrid carriers and allocating channels in the two-dimensional Fourier domain that yield optimal performance. The scheme enables spectral bandwidth and temporal bandwidth to be traded off, and provides flexibility in selecting demodulation strategies based on the features of the input. We present an in-depth comparison of different systems’ performances in various input features under the presence of noise. Simulation results show that the hybrid-modulation strategy offers the best comprehensive performance as compared to the conventional CSP and dual-scan techniques. © 2020 Optical Society of America
  Accession Number: 20204309399275
• Record 74 of
  
  Title:Evolution properties of the orbital angular momentum spectrum of twisted Gaussian Schell-model beams in turbulent atmosphere
  
  Author(s):Zhou, Mengyao(1); Fan, Weichen(2,3); Wu, Gaofeng(1)
  Source: Journal of the Optical Society of America A: Optics and Image Science, and Vision  Volume: 37  Issue: 1  DOI: 10.1364/JOSAA.37.000142  Published: 2020  
  Abstract:We derive the analytical formula of the energy weight of each orbital angular momentum (OAM) mode of twisted Gaussian Schell-model (TGSM) beams propagating in weak turbulent atmosphere. The evolution of its OAM spectrum is studied by numerical calculation. Our results show that the OAM spectrum of a TGSM beam changes with the beam propagating in turbulent atmosphere, which is completely different from that of the TGSM beam propagating in free space. Furthermore, influences of the source parameters and the turbulence parameters on the OAM spectrum of a TGSM beam in turbulent atmosphere are analyzed. It is found that the source parameters and turbulence parameters, such as twist factor, coherence length, beam waist size, and structure constant, have a significant influence on the OAM spectrum, but the value of the wavelength and inner scale have little influence. Increasing the beam waist size or decreasing the coherence length would lead to the OAM spectrum broadened in the source plane, but would be robust for the OAM modes of the TGSM beam in the turbulent atmosphere. It is clear that the bigger the value of the twist factor, the more asymmetric the OAM mode of the TGSM beam is, and the better mode distribution can be maintained when it propagates in turbulent atmosphere. Our results have potential applications in reducing the error rate of free-space optical communication and detecting the atmospheric parameters. © 2019 Optical Society of America
  Accession Number: 20200207984037
• Record 75 of
  
  Title:Combining competitive sequestration with nonlinear hybridization chain reaction amplification: an ultra-specific and highly sensitive sensing strategy for single-nucleotide variants
  
  Author(s):Zhao, Yan(1); Feng, Yuanbo(1); Zhang, Yuanbo(1); Xia, Pu(2); Xiao, Zihan(3); Wang, Ziheng(3); Yan, Hongxia(1)
  Source: Analytica Chimica Acta  Volume: 1130  Issue:   DOI: 10.1016/j.aca.2020.07.022  Published: 15 September 2020  
  Abstract:Highly specific and sensitive detection of single-nucleotide variants (SNVs) is of central importance in disease diagnosis and pharmacogenomics. However, it remains a great challenge to successfully detect very low amounts of mutant SNV sequences in real samples in which a SNV sequence may be surrounded by high levels of closely related wild-type sequences. Herein, we propose an ultra-specific and highly sensitive SNV sensing strategy by combining the competitive sequestration with the nonlinear hybridization chain reaction (HCR) amplification. The rationally designed sequestration hairpin can effectively sequester the large amount of wild-type sequence and thus dramatically improve the hybridization specificity in recognizing SNVs. To improve the detection sensitivity, a new fluorescent signal probe is fabricated by intercalating SYBR Green I dye into the nonlinear HCR based DNA dendrimer to further bind with SNVs for signal amplification. The hyperbranched DNA dendrimer possesses large numbers of DNA duplexes for dye intercalation, thus the signal probe shows strong fluorescence intensity, leading to large fluorescence signal amplification. Taking advantage of the improved hybridization specificity of the competitive sequestration and the enhanced fluorescence response of the nonlinear HCR amplification, the developed sensing strategy enables ultra-specific and highly sensitive detection of SNVs. Taking human pancreatic cancers and colorectal carcinomas related KRAS gene mutations as models, the developed strategy shows remarkably high specificity against 17 SNVs (discrimination factors ranged from 126 to 1001 with a median of 310), and achieves high sensitivity for 6 KRAS mutations (the best resultant detection limit reached 15 pM for KRAS G13D (c.38G > A)). Notably, combined with PCR amplification, our SNV sensing strategy could detect KRAS G12D (c.35G > A) from extracted human genomic DNA samples at abundance as low as 0.05%. This work expands the rule set of designing specific and sensitive SNV sensing strategies and shows promising potential application in clinical diagnosis. © 2020 Elsevier B.V.
  Accession Number: 20203309036239
• Record 76 of
  
  Title:Learning blur invariant binary descriptor for face recognition
  
  Author(s):Zhao, Chen(1,2); Li, Xuelong(3); Dong, Yongsheng(3)
  Source: Neurocomputing  Volume: 404  Issue:   DOI: 10.1016/j.neucom.2020.04.082  Published: 3 September 2020  
  Abstract:Binary representations have demonstrated remarkable performance in face recognition for its robustness to local changes and computation efficiency. However, the performance of face recognition based on most binary descriptors are not satisfactory when dealing with blurred face images. To solve this problem, we propose a novel blur invariant binary descriptor for face recognition. Particularly, we maximize the correlation between the binary codes of sharp face images and blurred face images of positive image pairs for learning the projection matrix. After that, we use the learned projection matrix to obtain blur-robust binary codes by quantizing projected pixel difference vectors (PDVs) in the testing stage. Experiment results on FERET and CMU-PIE show that our method achieves better recognition performance than representative binary descriptors LBP and CBFD. © 2020 Elsevier B.V.
  Accession Number: 20202108691276
• Record 77 of
  
  Title:A Survey of Human Action Analysis in HRI Applications
  
  Author(s):Ji, Yanli(1); Yang, Yang(1); Shen, Fumin(1); Shen, Heng Tao(1); Li, Xuelong(2)
  Source: IEEE Transactions on Circuits and Systems for Video Technology  Volume: 30  Issue: 7  DOI: 10.1109/TCSVT.2019.2912988  Published: July 2020  
  Abstract:The human action is an important information source for human social interaction, and it simultaneously plays a crucial role in human-robot interaction (HRI). For a natural and fluent interaction, robots are required to understand human actions and have the capacity to predict action intentions and to imitate human actions for an appropriate response. Currently, existing survey papers for the action recognition mainly summarize algorithms that perform action recognition in experimental scenarios, and survey papers of the HRI mainly introduced various interaction interfaces in the HRI. Different from these surveys, we focus on the human action analysis on robot platforms for the HRI application, including the body motion and gestures. We review the existing HRI related references involving the action recognition, prediction, and the robot imitation of the human action. Moreover, we give a summary of robot platforms and action datasets that are frequently used in the study of HRI. Finally, we give an analysis on the development trend and future research directions of action analysis for the HRI applications. © 1991-2012 IEEE.
  Accession Number: 20202908946279
• Record 78 of
  
  Title:Nonlocal Graph Convolutional Networks for Hyperspectral Image Classification
  
  Author(s):Mou, Lichao(1); Lu, Xiaoqiang(2); Li, Xuelong(3); Zhu, Xiao Xiang(1)
  Source: IEEE Transactions on Geoscience and Remote Sensing  Volume: 58  Issue: 12  DOI: 10.1109/TGRS.2020.2973363  Published: December 2020  
  Abstract:Over the past few years making use of deep networks, including convolutional neural networks (CNNs) and recurrent neural networks (RNNs), classifying hyperspectral images has progressed significantly and gained increasing attention. In spite of being successful, these networks need an adequate supply of labeled training instances for supervised learning, which, however, is quite costly to collect. On the other hand, unlabeled data can be accessed in almost arbitrary amounts. Hence it would be conceptually of great interest to explore networks that are able to exploit labeled and unlabeled data simultaneously for hyperspectral image classification. In this article, we propose a novel graph-based semisupervised network called nonlocal graph convolutional network (nonlocal GCN). Unlike existing CNNs and RNNs that receive pixels or patches of a hyperspectral image as inputs, this network takes the whole image (including both labeled and unlabeled data) in. More specifically, a nonlocal graph is first calculated. Given this graph representation, a couple of graph convolutional layers are used to extract features. Finally, the semisupervised learning of the network is done by using a cross-entropy error over all labeled instances. Note that the nonlocal GCN is end-to-end trainable. We demonstrate in extensive experiments that compared with state-of-the-art spectral classifiers and spectral-spatial classification networks, the nonlocal GCN is able to offer competitive results and high-quality classification maps (with fine boundaries and without noisy scattered points of misclassification). © 1980-2012 IEEE.
  Accession Number: 20205009606065
• Record 79 of
  
  Title:Design of a femtosecond electron diffractometer with adjustable gaps
  
  Author(s):Luo, Duan(1,2,3); Hui, Dan-Dan(1,2); Wen, Wen-Long(1); Li, Li-Li(1,2,3); Xin, Li-Wei(1); Zhong, Zi-Yuan(1,2,3); Ji, Chao(1,2,3); Chen, Ping(1); He, Kai(1); Wang, Xing(1); Tian, Jin-Shou(1,3)
  Source: Wuli Xuebao/Acta Physica Sinica  Volume: 69  Issue: 5  DOI: 10.7498/aps.69.20191157  Published: March 5, 2020  
  Abstract:One of the grand challenges in ultrafast science is real-time visualization of the microscopic structural evolution on atomic time and length scales. A promising pump-probe technique using a femtosecond laser pulse to initiate the ultrafast dynamics and another ultrashort electron pulse to probe the resulting changes has been developed and widely used to study ultrafast structural dynamics in chemical reactions, phase transitions, charge density waves, and even biological functions. In the past three decades, a number of different ultrafast electron guns have been developed to generate ultashort electron sources, mainly including hybrid electron gun with radio-frequency (RF) cavities for compressing the pulse broadening, relativistic electron gun for suppressing the coulomb interaction, single-electron pulses without space charge effect and compact direct current (DC) electron gun for minimizing the electron propagation distance. At present, these developments with different final electron energy and available total charge have improved the time response of ultrafast electron diffraction (UED) setups to a new frontier approaching to 100 fs regime. Although enormous efforts have been made, the superior capabilities and potentials of ultrafast electron diffraction (UED) are still hindered by space-charge induced pulse broadening. Besides, the penetration depth of electrons increases with the electron energy, while the scattering probability of electrons has the opposite consequence. Thus, in addition to the temporal resolution enhancement, it is also important that the electron energy should be tunable in a wide range to meet the requirements for samples with different thickness. Here in this work, we design a novel ultra-compact electron gun which combines a well-designed cathode profile, thereby providing a uniform field and a movable anode configuration to achieve a temporal resolution on the order of 100 fs over an accelerating voltage range from 10 kV to 125 kV. By optimizing the design of the high-voltage electrode profile, the field enhancement factor on the axis and along the cathode surface are both less than ~4% at different cathode-anode spacings, and thus the maximum on-axis field strength of ~10 MV/m is achieved under various accelerating voltages. This effectively suppresses the space charge broadening effect of the electron pulse. Furthermore, the anode aperture is designed as a stepped hole in which the dense sample grid can be placed, and the sample under study is directly supported by the grid and located at the anode, which reduces the cathode-to-sample distance, thus minimizing the electron pulse broadening from the cathode to sample. Moreover, the defocusing effect caused by the anode hole on the electron beam can be effectively reduced, therefore improving the lateral focusing performance of the electron beam. © 2020 Chinese Physical Society.
  Accession Number: 20202108693870
• Record 80 of
  
  Title:Theory and Method of Fourier Transform Hyperspectral Mueller Matrix Imaging
  
  Author(s):Liu, Jie(1); Li, Jianxin(1); Bai, Caixun(2); Xu, Yixuan(1); Qian, Jiamin(1); Wang, Yubo(1)
  Source: Guangxue Xuebao/Acta Optica Sinica  Volume: 40  Issue: 7  DOI: 10.3788/AOS202040.0711004  Published: April 10, 2020  
  Abstract:A hyperspectral Mueller matrix imaging (HMMI) method to capture spatial, spectral and Mueller matrix images at the same time is proposed. The principle of hyperspectral Mueller matrix imaging and the shear interference imaging process of birefringent interferometer are discussed in detail. The joint optimization design of polarization state generator and polarization state analyzer as well as the calibration method of this system is shown. In order to verify the performance of the instrument, the spectral Mueller matrix imaging of the target in the laboratory proves the feasibility of HMMI in quickly acquiring spectral images and Mueller matrix images. Because of its high spectral resolution and fast polarization modulation, it provides a new idea for the development of spectral Mueller matrix imaging. © 2020, Chinese Lasers Press. All right reserved.
  Accession Number: 20202308782384
• Record 81 of
  
  Title:Research and application of spectral reconstruction technology based on periodic structure
  
  Author(s):Liu, Bin(1,2); Wei, Ru Yi(1,3); Shi, Yi Shi(2,3); Shi, Lei(1,2); Zhang, Zai Kun(1,2); Zhao, Lv Rong(1,2); Zhang, Xin Ming(1,2)
  Source: Proceedings of SPIE - The International Society for Optical Engineering  Volume: 11567  Issue:   DOI: 10.1117/12.2579501  Published: 2020  
  Abstract:With the progress of science and the development of society, the study of material composition becomes more and more important, and the identification of material composition is mainly to distinguish the spectral information of different substances. Spectrometer is an important optical instrument, which combines the optical method with modern electronic data processing technology, and accurately analyzes the structure, composition and content of the target substance by obtaining its spectral information. At present, it has been widely used in some important scientific fields such as field exploration, material detection and space-borne analysis. In this paper, a new type of spectrometer based on periodic array structure is proposed. The spectrometer modulates the phase of the incident light by using the small hole array structure with the diameter of sub-wavelength. By using the diffraction effect of light, the light intensity distribution of the incident light is directly recorded by Charge Coupled Device1/4 CCD1/4according to different wavelengths. After passing through the diffraction aperture array, the light intensity distribution is recorded again, and the transmittance coefficients of the diffraction aperture array for different wavelengths are obtained respectively. Finally, the transmittance matrix of diffraction aperture array for incident light is obtained, and the spectral curves of different incident light can be obtained by data processing algorithm according to the transmittance matrix. The main optical device of the spectrometer is diffraction aperture array, which is a metal film coated on a transparent substrate made of resin material. A series of aperture arrays with diameters of 2-78 microns and 10*10 are processed on the metal film by micro-nano processing technology. Any aperture has different diameters, and the apertures are periodically arranged on the metal film. By using diffraction effect, the incident light with different wavelengths will produce different light intensity distribution on CCD. Combined with data processing method, the incident light can be obtained. Compared with the traditional spectrometer, the new spectrometer has no moving parts in the system, which improves its stability, and has the characteristics of fast data processing, small size, low cost and high spectral resolution. In this paper, the theoretical analysis, simulation and experimental verification of the new spectrometer are carried out. The results show that the new spectrometer has obvious advantages compared with the traditional spectrometer and has broad application prospects. © 2020 SPIE.
  Accession Number: 20205009602489
• Record 82 of
  
  Title:Supervised deep hashing with a joint deep network
  
  Author(s):Chen, Yaxiong(1,2); Lu, Xiaoqiang(1); Li, Xuelong(3)
  Source: Pattern Recognition  Volume: 105  Issue:   DOI: 10.1016/j.patcog.2020.107368  Published: September 2020  
  Abstract:Hashing has gained great attention in large-scale image retrieval due to efficient storage and fast search. Recently, many deep hashing approaches have achieved good results since deep neural network owns powerful learning capability. However, these deep hashing approaches can perform deep features learning and binary-like codes learning synchronously, the information loss between binary-like codes and binary codes will increase due to the binarization operation. A further deficiency is that binary-like codes learning based on deep feature representations is a shallow learning procedure, which cannot fully exploit deep feature representations to generate hash codes. To solve the above problems, we propose a Deep Learning Supervised Hashing (DLSH) method which adopts deep structure to learn binary codes based on deep feature representations for large-scale image retrieval. Specifically, we integrate deep features learning module, deep mapping module and binary codes learning module in one unified architecture. The network is trained in an end-to-end way. In addition, a new objective function is designed to preserve the balancing property and semantic similarity of binary codes by incorporating the semantic similarity term and the balanceable property term. Experimental results on four benchmarks demonstrate that the proposed approach outperforms several state-of-the-art hashing methods. © 2020
  Accession Number: 20203409068416
• Record 83 of
  
  Title:Method to control near-field bowing of laser diode arrays by balancing the thermal-induced stress
  
  Author(s):Zhang, Hongyou(1,2,3); Zah, Chung-En(3); Liu, Xingsheng(1,2,3,4)
  Source: Optical Engineering  Volume: 59  Issue: 3  DOI: 10.1117/1.OE.59.3.036104  Published: March 1, 2020  
  Abstract:Due to the thermal-induced stress during the bonding process, the emitters in a laser diode array (LDA) are vertically displaced, which causes the near-field bowing of a laser diode bar (i.e., the SMILE effect). Near-field bowing degrades the laser beam brightness, adversely affecting optical coupling and beam shaping, resulting in a larger divergence angle and a wider line after focusing and collimation. The mechanism of near-field bowing has been theoretically studied, in which the ratio of tensile strength between submount and heat sink has a great effect on the deformation of LDAs. Arm-wrestling between CuW submount and heat sink vividly describes that the deformation of LDAs changes as a function of the ratio of two materials' tensile strength. We design a symmetrical structure that bonds another submount on the bottom of the heat sink to control the SMILE effect by balancing the acting force from the top of the heat sink. The deformation of the heat sink and LDAs are approximately zero when the thermal-induced stresses forced on the top and bottom of the heat sink are equal. © 2020 Society of Photo-Optical Instrumentation Engineers (SPIE).
  Accession Number: 20201608414373
• Record 84 of
  
  Title:The laser-induced damage change detection for optical elements using siamese convolutional neural networks
  
  Author(s):Kou, Jingwei(1,2); Zhan, Tao(3); Zhou, Deyun(1); Wang, Wei(2); Da, Zhengshang(2); Gong, Maoguo(3)
  Source: Applied Soft Computing Journal  Volume: 87  Issue:   DOI: 10.1016/j.asoc.2019.106015  Published: February 2020  
  Abstract:Due to the fact that weak and fake laser-induced damages may occur in the surface of optical elements in high-energy laser facilities, it is still a challenging issue to effectively detect the real laser-induced damage changes of optical elements in optical images. Different from the traditional methods, in this paper, we put forward a similarity metric optimization driven supervised learning model to perform the laser-induced damage change detection task. In the proposed model, an end-to-end siamese convolutional neural network is designed and trained which can integrate the difference image generating and difference image analysis into a whole network. Thus, the damage changes can be highlighted by the pre-trained siamese network that classifies the central pixel between input multi-temporal image patches into changed and unchanged classes. To address the problem of unbalanced distribution between positive and negative samples, a modified average frequency balancing based weighted softmax loss is used to train the proposed network. Experiments conducted on two real datasets demonstrate the effectiveness and superiority of the proposed model. © 2019 Elsevier B.V.
  Accession Number: 20195207922532