2016

• Record 373 of
  
  Title:Non-uniform sampling knife-edge method for camera modulation transfer function measurement
  
  Author(s):Duan, Yaxuan(1,2); Xue, Xun(1); Chen, Yongquan(1); Tian, Liude(1,2); Zhao, Jianke(1); Gao, Limin(1)
  Source: Proceedings of SPIE - The International Society for Optical Engineering  Volume: 10023  Issue:   DOI: 10.1117/12.2245840  Published: 2016  
  Abstract:Traditional slanted knife-edge method experiences large errors in the camera modulation transfer function (MTF) due to tilt angle error in the knife-edge resulting in non-uniform sampling of the edge spread function. In order to resolve this problem, a non -uniform sampling knife-edge method for camera MTF measurement is proposed. By applying a simple direct calculation of the Fourier transform of the derivative for the non-uniform sampling data, the camera super-sampled MTF results are obtained. Theoretical simulations for images with and without noise under different tilt angle errors are run using the proposed method. It is demonstrated that the MTF results are insensitive to tilt angle errors. To verify the accuracy of the proposed method, an experimental setup for camera MTF measurement is established. Measurement results show that the proposed method is superior to traditional methods, and improves the universality of the slanted knife-edge method for camera MTF measurement. © 2016 SPIE.
  Accession Number: 20170603327553
• Record 374 of
  
  Title:Image de-fencing with hyperspectral camera
  
  Author(s):Zhang, Qi(1,2); Yuan, Yuan(1); Lu, Xiaoqiang(1)
  Source: IEEE CITS 2016 - 2016 International Conference on Computer, Information and Telecommunication Systems  Volume:   Issue:   DOI: 10.1109/CITS.2016.7546396  Published: August 16, 2016  
  Abstract:The main idea of image de-fencing refers to removing fence-like obstacles in the image and recovering the image. In this paper, rather than using a common RGB camera, we propose a novel image de-fencing algorithm with the help of a hyperspectral camera. Our algorithm consists of two phases: (1) automatically finding the location of the fence in the image, (2) image inpainting to reveal a fence-free image. With a hyperspectral camera, hundreds of images of the same scene under different wavelengths can be obtained instantly. By exploiting the spectral information of different positions in the scene with these hyperspectral images, the location of the fence can be distinguished from other objects. Then the fence can be removed and the image can be recovered with a novel image inpainting algorithm based on an approximate near-neighbor search method. Experiments demonstrate that our algorithm achieves considerable performance for the image de-fencing problem. © 2016 IEEE.
  Accession Number: 20163802815456
• Record 375 of
  
  Title:Unsupervised feature selection with structured graph optimization
  
  Author(s):Nie, Feiping(1); Zhu, Wei(1); Li, Xuelong(2)
  Source: 30th AAAI Conference on Artificial Intelligence, AAAI 2016  Volume:   Issue:   DOI:   Published: 2016  
  Abstract:Since amounts of unlabelled and high-dimensional data needed to be processed, unsupervised feature selection has become an important and challenging problem in machine learning. Conventional embedded unsupervised methods always need to construct the similarity matrix, which makes the selected features highly depend on the learned structure. However real world data always contain lots of noise samples and features that make the similarity matrix obtained by original data can't be fully relied. We propose an unsupervised feature selection approach which performs feature selection and local structure learning simultaneously, the similarity matrix thus can be determined adaptively. Moreover, we constrain the similarity matrix to make it contain more accurate information of data structure, thus the proposed approach can select more valuable features. An efficient and simple algorithm is derived to optimize the problem. Experiments on various benchmark data sets, including handwritten digit data, face image data and biomedical data, validate the effectiveness of the proposed approach. © 2016, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.
  Accession Number: 20165203195386
• Record 376 of
  
  Title:Far-field focal spot measurement of 10kJ-level laser facility
  
  Author(s):Wang, Zheng-Zhou(1,3,4); Xia, Yan-Wen(2); Li, Hong-Guang(4); Hu, Bing-Liang(4); Yin, Qin-Ye(1); Zheng, Kui-Xing(2)
  Source: Guangzi Xuebao/Acta Photonica Sinica  Volume: 45  Issue: 8  DOI: 10.3788/gzxb20164508.0812001  Published: August 1, 2016  
  Abstract:In order to evaluate the far-field beam quality of 10 kJ-level laser facility with different off-axis wedged focus lens, by utilizing the methods of the sampling of weak light beams and amplification imaging of splitting beams, the focal spot data of 3ω laser was collected by two 16-bit scientific-grade CCD cameras in the paths of main lobe and side lobe under the conditions of that the lateral magnification coefficient is the same but the intensity attenuation coefficient is different. One CCD obtained main lobe of far-field image, the other acquired its side lobe. The far-field focal spot was reconstructed based on the mathematical model of schlieren method, and the dynamic range is 1 151.7∶1. The influence of CCD dynamic range, relative magnification ratio and system noise on reconstructed image was analyzed. Experimental results show that, the method can achieve a high dynamic range far-field accurate measurement of focal spot, the stitching error is less than one pixel, which meets the requirements of targeting experiments in experimental precision. © 2016, Science Press. All right reserved.
  Accession Number: 20163402737309
• Record 377 of
  
  Title:Deep object tracking with multi-modal data
  
  Author(s):Zhang, Xuezhi(1,2); Yuan, Yuan(1); Lu, Xiaoqiang(1)
  Source: IEEE CITS 2016 - 2016 International Conference on Computer, Information and Telecommunication Systems  Volume:   Issue:   DOI: 10.1109/CITS.2016.7546403  Published: August 16, 2016  
  Abstract:Object tracking is a challenging topic in the field of computer vision since its performance is easily disturbed by occlusion, illumination change, background clutter, scale variation, etc. In this paper, we introduce a robust tracking algorithm that fuses information from both visible images and infrared (IR) images. The proposed tracking algorithm not only incorporates convolutional feature maps from the visible channel, but also employs a scale pyramid representation from IR channel. We estimate the target location by fusing multilayer convolutional feature maps, and predict the target scale from a scale pyramid. The pipeline of the proposed method is as follows. First, the hierarchical convolutional feature maps are obtained from visible images using VGG-Nets. Then, the accurate target location is predicted by the maximum response of correlation filters with the visible image feature maps. Finally, we obtain the precise object scale with a scale pyramid from infrared images where the difference between the target and the background is clear. In order to verify the performance of the proposed method, we capture six video sequences under different conditions. These sequences contain both visible channel and IR channel. Ten state-of-the-art tracking algorithms are compared with our method, and the experimental results show the effectiveness of the proposed tracker. © 2016 IEEE.
  Accession Number: 20163802815463
• Record 378 of
  
  Title:Robust object tracking via diverse templates
  
  Author(s):Wu, Siyuan(1,2); Li, Xuelong(1); Lu, Xiaoqiang(1)
  Source: IEEE CITS 2016 - 2016 International Conference on Computer, Information and Telecommunication Systems  Volume:   Issue:   DOI: 10.1109/CITS.2016.7546394  Published: August 16, 2016  
  Abstract:Robust object tracking is a challenging task in computer vision. Since the appearance of the target changes frequently, how to build and update the appearance model is crucial. In this paper, to better represent the object dynamically, we propose a robust object tracker based on diverse templates. First, we construct diverse multiple templates using the determinantal point process algorithm adaptively, which efficiently detects the most diverse subset of a set. Second, a patch-matching method is employed to propagate every template density to the next frame, and a voting map for each template is constructed by all matching patches. Third, a weighted Bayesian filter framework aggregates all voting maps to optimize target state. Finally, in order to maintain the diversity of multiple templates, we dynamically add, remove and replace the target from templates. Experimental results prove that the proposed method outperforms state-of-the-art tracking algorithms significantly in terms of center position errors and success rates. © 2016 IEEE.
  Accession Number: 20163802815454
• Record 379 of
  
  Title:Guest Editorial Special Section on Learning in Non-(geo)metric Spaces
  
  Author(s):Pelillo, Marcello(1); Hancock, Edwin R.(2); Li, Xuelong(3); Murino, Vittorio(4)
  Source: IEEE Transactions on Neural Networks and Learning Systems  Volume: 27  Issue: 6  DOI: 10.1109/TNNLS.2016.2522770  Published: June 2016  
  Abstract:Traditional machine learning and pattern recognition techniques are intimately linked to the notion of feature spaces. Adopting this view, each object is described in terms of a vector of numerical attributes and is, therefore, mapped to a point in a Euclidean (geometric) vector space, so that the distances between the points reflect the observed (dis)similarities between the respective objects. This kind of representation is attractive because geometric spaces offer powerful analytical as well as computational tools that are simply not available in other representations. Indeed, classical machine learning methods are tightly related to geometrical concepts, and numerous powerful tools have been developed during the last few decades, starting from the maximal likelihood method in the 1920s to perceptrons in the 1960s and, more recently, to kernel machines and deep learning architectures. © 2012 IEEE.
  Accession Number: 20162402481827
• Record 380 of
  
  Title:A new strategy lung nodules detection algorithm
  
  Author(s):Qiu, Shi(1,2); Wen, De-Sheng(1); Feng, Jun(3); Cui, Ying(4)
  Source: Tien Tzu Hsueh Pao/Acta Electronica Sinica  Volume: 44  Issue: 6  DOI: 10.3969/j.issn.0372-2112.2016.06.023  Published: June 1, 2016  
  Abstract:When lung nodules are detected in lung CT by computers,the vessel cross section and lung nodule have similar imaging characteristics in the two-dimensional CT image sequence,resulting in unable to detect problems precisely.We employed a new strategy for the lung nodules detection algorithm,which is based on the Gestalt psychology.This method can detect lung nodules indirectly by removing blood vessels.The experimental results show that,this algorithm can effectively reduce the influence of blood vessels on lung nodule detection,so as to improve the accuracy of detection of lung nodules. © 2016, Chinese Institute of Electronics. All right reserved.
  Accession Number: 20163002637996
• Record 381 of
  
  Title:A novel spatial-spectral sparse representation for hyperspectral image classification based on neighborhood segmentation
  
  Author(s):Wang, Cai-Ling(1,2); Wang, Hong-Wei(3); Hu, Bing-Liang(1); Wen, Jia(4); Xu, Jun(5); Li, Xiang-Juan(2)
  Source: Guang Pu Xue Yu Guang Pu Fen Xi/Spectroscopy and Spectral Analysis  Volume: 36  Issue: 9  DOI: 10.3964/j.issn.1000-0593(2016)09-2919-06  Published: September 1, 2016  
  Abstract:Traditional hyperspectral image classification algorithms focus on spectral information application, however, with the increase of spatial resolution of hyperspectral remote sensing images, hyperspectral imaging presents clustering properties on spatial domain for the same category. It is critical for hyperspectral image classification algorithms to use spatial information in order to improve the classification accuracy. However, the marginal differences of different categories display more obviously. If it is introduced directly into the spatial-spectral sparse representation for image classification without the selection of neighborhood pixels, the classification error and the computation time will increase. This paper presents a spatial-spectral joint sparse representation classification algorithm based on neighborhood segmentation. The algorithm calculates the similarity with spectral angel in order to choose proper neighborhood pixel into spatial-spectral joint sparse representation model. With simultaneous subspace pursuit and simultaneous orthogonal matching pursuit to solve the model, the classification is determined by computing the minimum reconstruction error between testing samples and training pixels. Two typical hyperspectral images from AVIRIS and ROSIS are chosen for simulation experiment and results display that the classification accuracy of two images both improves as neighborhood segmentation threshold increasing. It concludes that neighborhood segmentation is necessary for joint sparse representation classification. © 2016, Peking University Press. All right reserved.
  Accession Number: 20163902850948
• Record 382 of
  
  Title:A 60GHz RoF(radio-over-fiber) transmission system based on PM modulator
  
  Author(s):Wang, Xin(1,2); Liu, Yi(3); Wang, Wen-Ting(2)
  Source: Proceedings of SPIE - The International Society for Optical Engineering  Volume: 10017  Issue:   DOI: 10.1117/12.2246651  Published: 2016  
  Abstract:As one of the most important applications of microwave photonic, ROF (Radio over Fiber) system, which combines the advantages of optical communication and wireless communication, is a good candidate for broadband mobile Communication In this paper, we built and simulation a 60GHz RoF(Radio-over-Fiber) transmission system based on PM modulator. First, we introduce the PM-IM(Phase modulation to intensity modulation) modulation mechanisms by the breaking the phase balanced approach. This method solves the problem that the constant envelope (phase modulation signal) generated by the phase modulator can not be directly detected by a photo detector. A standard single-mode fiber (SMF) is connected input to the F-P(Fabry-Perot) optical filter, which is to achieve the PM-IM modulation conversion by changing the wavelength of the laser or the frequency of the modulation factor of the F-P optical filter to adapt to different fiber lengths and the signal transmission rate. These two methods which changing the phase relationship between the optical carrier and the optical side band can realize the ideal phase transition to obtain efficient and low loss modulation conversion. Finally, the simulation results show that different fiber lengths and the signal transmission rate configuration of different wavelength of the laser or the frequency of the modulation factor of the F-P optical filter, the BER performance and the eye diagram of the 60GHz RoF transmission system signals have been improved based on these PM-IM modulation methods. © 2016 SPIE.
  Accession Number: 20170503309781
• Record 383 of
  
  Title:Ultra-high Q one-dimensional hybrid PhC-SPP waveguide microcavity with large structure tolerance
  
  Author(s):Liu, Feng(1); Zhang, Lingxuan(1,2,3); Lu, Xiaoyuan(1,3); Wang, Weiqiang(1); Wang, Leiran(1); Wang, Guoxi(1,2); Zhang, Wenfu(1,2); Zhao, Wei(1,2)
  Source: Journal of Modern Optics  Volume: 63  Issue: 12  DOI: 10.1080/09500340.2015.1130272  Published: July 3, 2016  
  Abstract:A photonic crystal - surface plasmon-polaritons hybrid transverse magnetic mode waveguide based on a one-dimensional optical microcavity is designed to work in the communication band. A Gaussian field distribution in a stepping heterojunction taper is designed by band engineering, and a silica layer compresses the mode field to the subwavelength scale. The designed microcavity possesses a resonant mode with a quality factor of 1609 and a modal volume of 0.01 cubic wavelength. The constant period and the large structure tolerance make it realizable by current processing techniques. © 2016 Taylor & Francis.
  Accession Number: 20160201781837
• Record 384 of
  
  Title:Impact of light polarization on the measurement of water particulate backscattering coefficient
  
  Author(s):Liu, Jia(1,2); Gong, Fang(1); He, Xian-Qiang(1); Zhu, Qian-Kun(1); Huang, Hai-Qing(1)
  Source: Guang Pu Xue Yu Guang Pu Fen Xi/Spectroscopy and Spectral Analysis  Volume: 36  Issue: 1  DOI: 10.3964/j.issn.1000-0593(2016)01-0031-07  Published: January 1, 2016  
  Abstract:Particulate backscattering coefficient is a main inherent optical properties (IOPs) of water, which is also a determining factor of ocean color and a basic parameter for inversion of satellite ocean color remote sensing. In-situ measurement with optical instruments is currently the main method for obtaining the particulate backscattering coefficient of water. Due to reflection and refraction by the mirrors in the instrument optical path, the emergent light source from the instrument may be partly polarized, thus to impact the measurement accuracy of water backscattering coefficient. At present, the light polarization of measuring instruments and its impact on the measurement accuracy of particulate backscattering coefficient are still poorly known. For this reason, taking a widely used backscattering coefficient measuring instrument HydroScat6 (HS-6) as an example in this paper, the polarization characteristic of the emergent light from the instrument was systematically measured, and further experimental study on the impact of the light polarization on the measurement accuracy of the particulate backscattering coefficient of water was carried out. The results show that the degree of polarization(DOP) of the central wavelength of emergent light ranges from 20% to 30% for all of the six channels of the HS-6, except the 590 nm channel from which the DOP of the emergent light is slightly low (~15%). Therefore, the emergent light from the HS-6 has significant polarization. Light polarization has non-neglectable impact on the measurement of particulate backscattering coefficient, and the impact degree varies with the wave band, linear polarization angle and suspended particulate matter(SPM) concentration. At different SPM concentrations, the mean difference caused by light polarization can reach 15.49%, 11.27%, 12.79%, 14.43%, 13.76%, and 12.46% in six bands, 420, 442, 470, 510, 590, and 670 nm, respectively. Consequently, the impact of light polarization on the measurement of particulate backscattering coefficient with an optical instrument should be taken into account, and the DOP of the emergent light should be reduced as much as possible. © 2016, Science Press. All right reserved.
  Accession Number: 20160101768426