2015

• Record 133 of
  
  Title:Blind image quality assessment via deep learning
  
  Author(s):Hou, Weilong(1); Gao, Xinbo(1); Tao, Dacheng(2); Li, Xuelong(3)
  Source: IEEE Transactions on Neural Networks and Learning Systems  Volume: 26  Issue: 6  DOI: 10.1109/TNNLS.2014.2336852  Published: June 1, 2015  
  Abstract:This paper investigates how to blindly evaluate the visual quality of an image by learning rules from linguistic descriptions. Extensive psychological evidence shows that humans prefer to conduct evaluations qualitatively rather than numerically. The qualitative evaluations are then converted into the numerical scores to fairly benchmark objective image quality assessment (IQA) metrics. Recently, lots of learning-based IQA models are proposed by analyzing the mapping from the images to numerical ratings. However, the learnt mapping can hardly be accurate enough because some information has been lost in such an irreversible conversion from the linguistic descriptions to numerical scores. In this paper, we propose a blind IQA model, which learns qualitative evaluations directly and outputs numerical scores for general utilization and fair comparison. Images are represented by natural scene statistics features. A discriminative deep model is trained to classify the features into five grades, corresponding to five explicit mental concepts, i.e., excellent, good, fair, poor, and bad. A newly designed quality pooling is then applied to convert the qualitative labels into scores. The classification framework is not only much more natural than the regression-based models, but also robust to the small sample size problem. Thorough experiments are conducted on popular databases to verify the model's effectiveness, efficiency, and robustness. © 2012 IEEE.
  Accession Number: 20152200894482
• Record 134 of
  
  Title:The transmission of polarized light of space attitude in quantum communication
  
  Author(s):Yang, Hai-Ma(1,2,3); Ma, Cai-Wen(2); Wang, Jian-Yu(4); Zhang, Liang(4); Liu, Jin(1); Huan, Yuan-Shen(1)
  Source: Guangzi Xuebao/Acta Photonica Sinica  Volume: 44  Issue: 12  DOI: 10.3788/gzxb20154412.1227002  Published: December 1, 2015  
  Abstract:By using the design of the orthogonal polarized light beacon, the single optical path transmission of space beacons gesture was achieved, which provied the conditions for the Satelite-Ground quantum optical link. The transmission characteristic of the polarization through the optical device, especially the coated device was analyzed. A simulation was done to analyze the outgoing beacon light under the condition of the different incident angles and rotation angles. The influence by the phase and reflectivity difference in the optical components was analyzed. A mathematical model of the measurement of polarization azimuth by using the Jones matrix was made to analyze the form of the Malus law in the elliptic polarized light incident. Three-dimension attitude can be obtained by a single Position Sensitive Detector sensor which can receive the beacon light, decouple the angle of polarization and the location of incident light. The experiment data shows that the system has the function of measuring three-dimension attitude of the beacon by a single Position Sensitive Detector sensor. This system provides a solution to the fields of the Satelite-Ground Optical Communication and the measurement of space geometry position. © 2015, Chinese Optical Society. All right reserved.
  Accession Number: 20160201786522
• Record 135 of
  
  Title:Structured-patch optimization for dense correspondence
  
  Author(s):Qin, Xiameng(1); Shen, Jianbing(1); Mao, Xiaoyang(2); Li, Xuelong(3); Jia, Yunde(1)
  Source: IEEE Transactions on Multimedia  Volume: 17  Issue: 3  DOI: 10.1109/TMM.2015.2395078  Published: March 1, 2015  
  Abstract:This paper presents a new method to compute the dense correspondences between two images by using the energy optimization and the structured patches. In terms of the property of the sparse feature and the principle that nearest sub-scenes and neighbors are much more similar, we design a new energy optimization to guide the dense matching process and find the reliable correspondences. The sparse features are also employed to design a new structure to describe the patches. Both transformation and deformation with the structured patches are considered and incorporated into an energy optimization framework. Thus, our algorithm can match the objects robustly in complicated scenes. Finally, a local refinement technique is proposed to solve the perturbation of the matched patches. Experimental results demonstrate that our method outperforms the state-of-the-art matching algorithms. © 2015 IEEE.
  Accession Number: 20150900578988
• Record 136 of
  
  Title:Facile synthesis of 3D reduced graphene oxide and its polyaniline composite for super capacitor application
  
  Author(s):Tang, Wei(1); Peng, Li(2); Yuan, Chunqiu(1); Wang, Jian(1); Mo, Shenbin(1); Zhao, Chunyan(1); Yu, Youhai(3); Min, Yonggang(1); Epstein, Arthur J.(4)
  Source: Synthetic Metals  Volume: 202  Issue:   DOI: 10.1016/j.synthmet.2015.01.031  Published: April 2015  
  Abstract:We propose a facile and environmentally-friendly strategy for fabricating three-dimensional (3D) reduced graphene oxide (3D-rGO) porous structure with one step hydrothermal method using glucose as the reducing agent and CaCO3 as the template. The reducing process was accompanied by the self-assembly of two-dimensional graphene sheets into a 3D hydrogel which entrapped CaCO3 particle into the graphene network. After the removal of CaCO3 particle, 3D-rGO with interconnected porous structure was obtained. The 3D-rGO was further composted with PANI nanowire. The structure and the property of 3D-rGO and 3D-rGO/PANI composite have been characterized by X-ray photoelectron spectroscopy, Fourier transform infrared spectroscopy, X-ray diffraction, scanning electron microscopy, transmission electron microscopy, cyclic voltammetry, galvanostatic charge-discharge test and electrochemical impedance spectroscopy. Electrochemical test reveals that the 3D-rGO/PANI has high capacitance performance of 243 F g-1 at current charge-discharge current density of 1 A g-1 and an excellent capacity retention rate of 86% after 1000 cycles. © 2015 Elsevier B.V. All rights reserved.
  Accession Number: 20150700517042
• Record 137 of
  
  Title:A real-time axial activeanti-drift device with high-precision
  
  Author(s):Huo, Ying-Dong(1,2); Cao, Bo(2); Yu, Bin(2); Chen, Dan-Ni(2,3); Niu, Han-Ben(2)
  Source: Wuli Xuebao/Acta Physica Sinica  Volume: 64  Issue: 2  DOI: 10.7498/aps.64.028701  Published: January 20, 2015  
  Abstract:In a fluorescent nano-resolution microscope based on single molecular localization, drift of focal plane will bring an additional deviation to the accuracy of single molecular localization. Consequently, this will reduce the final resolution of the reconstructed image and cause image degradation. Therefore, it is vital to control the system drift to a minimum level as much as possible. In recent years, the anti-drift ways emerged in endlessly. In this paper we made a systematic study aiming at the method in which optical measurement and negative feedback control are used. The basic principle and its implementation of the system are analyzed, and possible error is also evaluated. Finally, the precision of the system is tested experimentally. With this device, axial drift can be detected and corrected automatically in time, and the axial anti-drift accuracy as high as 9.93 nm can be achieved, which is one order higher than that of the existing commercial microscopies. © 2015 Chinese Physical Society.
  Accession Number: 20150600487599
• Record 138 of
  
  Title:Person reidentification by minimum classification error-based KISS metric learning
  
  Author(s):Tao, Dapeng(1); Jin, Lianwen(1); Wang, Yongfei(1); Li, Xuelong(2)
  Source: IEEE Transactions on Cybernetics  Volume: 45  Issue: 2  DOI: 10.1109/TCYB.2014.2323992  Published: February 1, 2015  
  Abstract:In recent years, person reidentification has received growing attention with the increasing popularity of intelligent video surveillance. This is because person reidentification is critical for human tracking with multiple cameras. Recently, keep it simple and straightforward (KISS) metric learning has been regarded as a top level algorithm for person reidentification. The covariance matrices of KISS are estimated by maximum likelihood (ML) estimation. It is known that discriminative learning based on the minimum classification error (MCE) is more reliable than classical ML estimation with the increasing of the number of training samples. When considering a small sample size problem, direct MCE KISS does not work well, because of the estimate error of small eigenvalues. Therefore, we further introduce the smoothing technique to improve the estimates of the small eigenvalues of a covariance matrix. Our new scheme is termed the minimum classification error-KISS (MCE-KISS). We conduct thorough validation experiments on the VIPeR and ETHZ datasets, which demonstrate the robustness and effectiveness of MCE-KISS for person reidentification. © 2013 IEEE.
  Accession Number: 20150400447475
• Record 139 of
  
  Title:Representative and diverse video summarization
  
  Author(s):Chen, Xiao(1,2); Li, Xuelong(1); Lu, Xiaoqiang(1)
  Source: 2015 IEEE China Summit and International Conference on Signal and Information Processing, ChinaSIP 2015 - Proceedings  Volume:   Issue:   DOI: 10.1109/ChinaSIP.2015.7230379  Published: August 31, 2015  
  Abstract:Video summarization usually refers to produce a summary preserving essential content of the original video. Many existing methods have been developed to select representative frames by a dictionary learning model, which have led to a state-of-The-Art performance. However, learning dictionary without considering relationship between samples of the original data space would lead to imprecise representation. To address this problem, in this paper, geometrical distribution information of samples is incorporated into the dictionary learning process. A graph based learning strategy is employed to draw the geometrical distribution information. Meanwhile, the diversity criteria is considered as important as representativeness, which can reduce redundant frames to be selected in final summary. Thus similarity measuring is imported to guarantee that a final summary contains diversity contents within the original video. The proposed method is validated on a challenging and widely used dataset, and state-of-The-Art performance is achieved in contrast to other methods. © 2015 IEEE.
  Accession Number: 20160701912145
• Record 140 of
  
  Title:Texture classification and retrieval using shearlets and linear regression
  
  Author(s):Dong, Yongsheng(1,2); Tao, Dacheng(2); Li, Xuelong(2); Ma, Jinwen(3); Pu, Jiexin(1)
  Source: IEEE Transactions on Cybernetics  Volume: 45  Issue: 3  DOI: 10.1109/TCYB.2014.2326059  Published: March 1, 2015  
  Abstract:Statistical modeling of wavelet subbands has frequently been used for image recognition and retrieval. However, traditional wavelets are unsuitable for use with images containing distributed discontinuities, such as edges. Shearlets are a newly developed extension of wavelets that are better suited to image characterization. Here, we propose novel texture classification and retrieval methods that model adjacent shearlet subband dependences using linear regression. For texture classification, we use two energy features to represent each shearlet subband in order to overcome the limitation that subband coefficients are complex numbers. Linear regression is used to model the features of adjacent subbands; the regression residuals are then used to define the distance from a test texture to a texture class. Texture retrieval consists of two processes: the first is based on statistics in contourlet domains, while the second is performed using a pseudo-feedback mechanism based on linear regression modeling of shearlet subband dependences. Comprehensive validation experiments performed on five large texture datasets reveal that the proposed classification and retrieval methods outperform the current state-of-the-art. © 2013 IEEE.
  Accession Number: 20150900578558
• Record 141 of
  
  Title:Soliton dynamics in a PT-symmetric optical lattice with a longitudinal potential barrier
  
  Author(s):Zhou, Keya(1,2); Wei, Tingting(1); Sun, Haipeng(1); He, Yingji(3); Liu, Shutian(1)
  Source: Optics Express  Volume: 23  Issue: 13  DOI: 10.1364/OE.23.016903  Published: June 29, 2015  
  Abstract:We present dynamics of spatial solitons propagating through a PT symmetric optical lattice with a longitudinal potential barrier. We find that a spatial soliton evolves a transverse drift motion after transmitting through the lattice barrier. The gain/loss coefficient of the PT symmetric potential barrier plays an essential role on such soliton dynamics. The bending angle of solitons depends on the lattice parameters including the modulation frequency, incident position, potential depth and the barrier length. Besides, solitons tend to gain a certain amount of energy from the barrier, which can also be tuned by barrier parameters. © 2015 Optical Society of America.
  Accession Number: 20153701275010
• Record 142 of
  
  Title:Transfer learning for visual categorization: A survey
  
  Author(s):Shao, Ling(1,2); Zhu, Fan(2); Li, Xuelong(3)
  Source: IEEE Transactions on Neural Networks and Learning Systems  Volume: 26  Issue: 5  DOI: 10.1109/TNNLS.2014.2330900  Published: May 1, 2015  
  Abstract:Regular machine learning and data mining techniques study the training data for future inferences under a major assumption that the future data are within the same feature space or have the same distribution as the training data. However, due to the limited availability of human labeled training data, training data that stay in the same feature space or have the same distribution as the future data cannot be guaranteed to be sufficient enough to avoid the over-fitting problem. In real-world applications, apart from data in the target domain, related data in a different domain can also be included to expand the availability of our prior knowledge about the target future data. Transfer learning addresses such cross-domain learning problems by extracting useful information from data in a related domain and transferring them for being used in target tasks. In recent years, with transfer learning being applied to visual categorization, some typical problems, e.g., view divergence in action recognition tasks and concept drifting in image classification tasks, can be efficiently solved. In this paper, we survey state-of-the-art transfer learning algorithms in visual categorization applications, such as object recognition, image classification, and human action recognition. © 2012 IEEE.
  Accession Number: 20151700778981
• Record 143 of
  
  Title:Enhanced properties of poly(vinyl alcohol) composite films with functionalized graphene
  
  Author(s):Mo, Shenbin(1); Peng, Li(2); Yuan, Chunqiu(1); Zhao, Chunyan(1); Tang, Wei(1); Ma, Cunliang(1); Shen, Jiaxin(1); Yang, Wenbin(2); Yu, Youhai(3); Min, Yong(1); Epstein, Arthur J.(4)
  Source: RSC Advances  Volume: 5  Issue: 118  DOI: 10.1039/c5ra15984a  Published: 2015  
  Abstract:Three types of poly(vinyl alcohol) (PVA) composite films containing graphene oxide (GO), reduced graphene oxide (RGO) and novel sulfonated graphene oxide (SRGO) as a filler were successfully prepared by a simple solution casting. The structure and properties of graphene-based PVA composites films were investigated. The results showed that the properties of the polymer composites films were sensitive to the structure of graphene. GO acted as the best reinforcing filler to enhance the mechanical property of PVA because it has many oxygen functional groups which could enhance the interfacial interactions through the formation of hydrogen bonds with PVA chains. The tensile strength and modulus of the resulting PVA/GO composites could reach 280 MPa and 13.5 GPa, respectively. RGO could improve the dielectric properties of PVA and the electrical conductivities were increased by ∼1011 orders of magnitude in the composites with 50 wt% of filler loadings as compared to that of neat PVA. SRGO could enhance the mechanical and dielectric properties of PVA simultaneously. The mechanical properties of PVA could be efficiently improved due to the strong interaction between the -SO3H groups on the SRGO sheets and PVA chains. The tensile strength and modulus of the resulting PVA/SRGO composites could reach 252 MPa and 8.5 GPa, respectively. Although the conductivity values of PVA/SRGO composites were less than those of the PVA/RGO composites, they were still increased by ∼1010 orders of magnitude in the composites with 50 wt% of filler loadings as compared to that of neat PVA. These results demonstrated that PVA films with enhancement in the mechanical and electronic properties can be fabricated with proper modified graphene. © 2015 The Royal Society of Chemistry.
  Accession Number: 20154801611316
• Record 144 of
  
  Title:Computer simulation for hybrid plenoptic camera super-resolution refocusing with focused and unfocused mode
  
  Author(s):Zhang, Wei(1,2); Guo, Xin(1); You, Suping(1); Yang, Bo(1); Wan, Xinjun(1)
  Source: Hongwai yu Jiguang Gongcheng/Infrared and Laser Engineering  Volume: 44  Issue: 11  DOI:   Published: November 25, 2015  
  Abstract:Light field is a representation of full four-dimensional radiance of rays in free space. Plenoptic camera is a kind of system which could obtain light field image. In typical plenoptic camera, the final spatial resolution of the image is limited by the numbers of the microlens of the array. The focused plenoptic camera could capture a light field with higher spatial resolution than the traditional approach, but the directional resolution will be decreased for trading. Two models were set up to emulate the 4D light field distribution in both the traditional plenoptic camera and the focused plenoptic camera respectively. The 4D light field images of the two kinds of plenoptic camera were simulated by the software ZEMAX. The differences of sampling methods of the two kinds of plenoptic camera were analyzed. A variable focal length microlens array was presumed to be used in plenoptic camera to implement both focused and unfocused light field imaging. Based on the recorded light field, the corresponding refocusing process was discussed then. The refocused images at different depth were calculated. A new method of enhancing the resolution of the refocused images by image fusion and super resolution theories was presented. A reconstructed all in-focus image with resolution of 3 times of traditional plenoptic camera and same depth of field was achieved finally. © 2015, Editorial Board of Journal of Infrared and Laser Engineering. All right reserved.
  Accession Number: 20160101761487