2013
2013
-
Record 37 of
Title:Transductive face sketch-photo synthesis
Author(s):Wang, Nannan(1); Tao, Dacheng(2); Gao, Xinbo(1); Li, Xuelong(3); Li, Jie(1)Source: IEEE Transactions on Neural Networks and Learning Systems Volume: 24 Issue: 9 DOI: 10.1109/TNNLS.2013.2258174 Published: 2013Abstract:Face sketch-photo synthesis plays a critical role in many applications, such as law enforcement and digital entertainment. Recently, many face sketch-photo synthesis methods have been proposed under the framework of inductive learning, and these have obtained promising performance. However, these inductive learning-based face sketch-photo synthesis methods may result in high losses for test samples, because inductive learning minimizes the empirical loss for training samples. This paper presents a novel transductive face sketch-photo synthesis method that incorporates the given test samples into the learning process and optimizes the performance on these test samples. In particular, it defines a probabilistic model to optimize both the reconstruction fidelity of the input photo (sketch) and the synthesis fidelity of the target output sketch (photo), and efficiently optimizes this probabilistic model by alternating optimization. The proposed transductive method significantly reduces the expected high loss and improves the synthesis performance for test samples. Experimental results on the Chinese University of Hong Kong face sketch data set demonstrate the effectiveness of the proposed method by comparing it with representative inductive learning-based face sketch-photo synthesis methods. ? 2012 IEEE.Accession Number: 20133516674501 -
Record 38 of
Title:Fast and accurate matrix completion via truncated nuclear norm regularization
Author(s):Hu, Yao(1); Zhang, Debing(1); Ye, Jieping(2); Li, Xuelong(3); He, Xiaofei(1)Source: IEEE Transactions on Pattern Analysis and Machine Intelligence Volume: 35 Issue: 9 DOI: 10.1109/TPAMI.2012.271 Published: 2013Abstract:Recovering a large matrix from a small subset of its entries is a challenging problem arising in many real applications, such as image inpainting and recommender systems. Many existing approaches formulate this problem as a general low-rank matrix approximation problem. Since the rank operator is nonconvex and discontinuous, most of the recent theoretical studies use the nuclear norm as a convex relaxation. One major limitation of the existing approaches based on nuclear norm minimization is that all the singular values are simultaneously minimized, and thus the rank may not be well approximated in practice. In this paper, we propose to achieve a better approximation to the rank of matrix by truncated nuclear norm, which is given by the nuclear norm subtracted by the sum of the largest few singular values. In addition, we develop a novel matrix completion algorithm by minimizing the Truncated Nuclear Norm. We further develop three efficient iterative procedures, TNNR-ADMM, TNNR-APGL, and TNNR-ADMMAP, to solve the optimization problem. TNNR-ADMM utilizes the alternating direction method of multipliers (ADMM), while TNNR-AGPL applies the accelerated proximal gradient line search method (APGL) for the final optimization. For TNNR-ADMMAP, we make use of an adaptive penalty according to a novel update rule for ADMM to achieve a faster convergence rate. Our empirical study shows encouraging results of the proposed algorithms in comparison to the state-of-the-art matrix completion algorithms on both synthetic and real visual datasets. ? 1979-2012 IEEE.Accession Number: 20133216579103 -
Record 39 of
Title:An adaptive and effective single image dehazing algorithm based on dark channel prior
Author(s):Zhu, Qingsong(1,2); Yang, Shuai(3); Heng, Pheng Ann(4); Li, Xuelong(5)Source: 2013 IEEE International Conference on Robotics and Biomimetics, ROBIO 2013 Volume: Issue: DOI: 10.1109/ROBIO.2013.6739728 Published: 2013Abstract:In this paper, we describe a novel and effective single image enhancement algorithm for haze image. As we observe that, the contrast and intensity of haze image after using dark channel prior approach will unavoidably tend to be lower than those of the real scene, we proposed a method using histogram specification to make an improvement on image after dark channel prior approach. We make a large number of experiment and find that, if dealing with a haze image with large background area and low contrast, dark channel prior result will become dark, also a general haze image after dark channel occurs different degree of anamorphose. We introduce an adaptive algorithm to repair the different kinds of anamorphose on the hazy image after dark channel prior. The experimental results shows that our method make the dehazing result more close to real scene. ? 2013 IEEE.Accession Number: 20141717632973 -
Record 40 of
Title:Far-field tunable nano-focusing based on metallic slits surrounded with nonlinear-variant widths and linear-variant depths of circular dielectric grating
Author(s):Cao, Peng-Fei(1); Cheng, Ling(1,2); Zhang, Xiao-Ping(1); Lu, Wei-Ping(2); Kong, Wei-Jie(1,3); Liang, Xue-Wu(1,3)Source: Progress in Electromagnetics Research Volume: 138 Issue: DOI: 10.2528/PIER13011904 Published: 2013Abstract:In this work, we present a new design of a tunable nanofocusing lens using a circular grating of linear-variant depths and nonlinear-variant widths. Constructive interference of cylindrical surface plasmon launched by the subwavelength metallic structure forms a subdiffraction-limited focus, the focal length can be adjusted by varying the geometry of each groove in the circular grating. According to the numerical calculation, the range of focusing points shift is much more than other plasmonic lens, and the relative phase of emitting light scattered by surface plasmon coupling circular grating can be modulated by the nonlinear-variant width and linear-variant depth. The simulation result indicates that the different relative phase of emitting light lead to variant focal length. We firstly show a unique phenomenon for the linear-variant depths and nonlinear-variant widths of the circular grating that the positive change and negative change of the depths and widths of grooves can result in different of variation trend between relative phases and focal lengths. These results paved the road for utilizing the plasmonic lens in high-density optical storage, nanolithography, superresolution optical microscopic imaging, optical trapping, and sensing.Accession Number: 20131716230381 -
Record 41 of
Title:A novel rain detection and removal approach using guided filtering and formation modeling
Author(s):Zhu, Qingsong(1); Shao, Ling(2); Heng, Pheng Ann(3); Li, Xuelong(4)Source: 2013 IEEE International Conference on Robotics and Biomimetics, ROBIO 2013 Volume: Issue: DOI: 10.1109/ROBIO.2013.6739519 Published: 2013Abstract:The task of removing rain is of great significance for outdoor vision systems such as video surveillance, vision based navigation and so on. Rain produces complex time varying intensity fluctuations in images or videos, which seriously reduce the performance of outdoor vision systems. Due to similar visual appearances of rain and moving objects, the current rain removal algorithms cannot easily distinguish between the two. In this paper, we propose a novel algorithm for rain detection and removal based on the rain image formation model and edge-preserving filtering. The effectiveness of our algorithm is demonstrated in comparison with the existing approaches, by experimenting on videos of intricate scenes with moving objects or time-varying textures. ? 2013 IEEE.Accession Number: 20141717632768 -
Record 42 of
Title:A novel segmentation guided approach for single image dehazing
Author(s):Zhu, Qingsong(1,2); Heng, Pheng Ann(3); Shao, Ling(4); Li, Xuelong(5)Source: 2013 IEEE International Conference on Robotics and Biomimetics, ROBIO 2013 Volume: Issue: DOI: 10.1109/ROBIO.2013.6739832 Published: 2013Abstract:This paper presents a novel image haze removal approach from single image. In the algorithm, the constant albedo and dark channel prior methods are combined to represent the transmission model of hazed image. And then, the watershed segmentation approach is introduced to decompose the input image into some gray level consistent areas. Compared with traditional fixed image partition schemes, better estimation of the atmospheric light can be obtained as well as to avoid the problem of halo artifacts. With the improved haze image modeling approach and atmospheric light estimation, the dehazed image with better visual quality can be achieved. ? 2013 IEEE.Accession Number: 20141717633076 -
Record 43 of
Title:Systemic optimization of linear cavity Yb-doped double-clad fiber laser
Author(s):Liu, Jinglin(1); Zhao, Chujun(2); Hu, Huanlong(1); Shuai, Cijun(1,3)Source: Optik Volume: 124 Issue: 9 DOI: 10.1016/j.ijleo.2012.02.007 Published: May 2013Abstract:In order to optimize the double-clad fiber laser system, the laser output power should be as large as possible and the systemic cost should be as cheap as possible under the condition of the optimum fiber length. In this work, the improved approximate analytical solutions of the optimum fiber length and the laser output power are obtained based on the model of linear cavity strongly pumped Yb-doped double-clad fiber laser. And the effects of the laser scattering loss, the pump power, Yb dopant concentration, and the reflection coefficients of the input and output mirrors at laser wavelength on the laser output power and the optimum fiber length are discussed and analyzed in detail. Thereby the optimal laser system can be determined. ? 2012 Elsevier GmbH.Accession Number: 20131316140690 -
Record 44 of
Title:What's the role of image matting in image segmentation?
Author(s):Zhu, Qingsong(1,2); Heng, Pheng Ann(3); Shao, Ling(4); Li, Xuelong(5)Source: 2013 IEEE International Conference on Robotics and Biomimetics, ROBIO 2013 Volume: Issue: DOI: 10.1109/ROBIO.2013.6739711 Published: 2013Abstract:Image Matting is the key technology in image processing, video editing, and film-making applications. With the fast development of modern information technology, image matting has gained increasing interests from both academic and industrial communities. So what is Image Matting? And, What's the Role of Image Matting in Image Segmentation? In this paper, we will try to give a comprehensive and constructive answer to the above questions. ? 2013 IEEE.Accession Number: 20141717632956 -
Record 45 of
Title:Image encryption algorithm by using fractional Fourier transform and pixel scrambling operation based on double random phase encoding
Author(s):Liu, Zhengjun(1,2); Li, She(3); Liu, Wei(3); Wang, Yanhua(4); Liu, Shutian(3)Source: Optics and Lasers in Engineering Volume: 51 Issue: 1 DOI: 10.1016/j.optlaseng.2012.08.004 Published: January 2013Abstract:To enhance the security of double random phase encoding, a kind of amplitude scrambling operation is designed and introduced into an image encryption process. The random data of the second phase mask in double random phase encoding is also employed for scrambling amplitude distribution in order to save the space of storage and transmission of the key information. The scrambling operator is changeable for generating the key. Some numerical simulations have been provided for testing the validity of the image encryption scheme. ? 2012 Elsevier Ltd.Accession Number: 20124015490647 -
Record 46 of
Title:Hessian regularized support vector machines for mobile image annotation on the cloud
Author(s):Tao, Dapeng(1); Jin, Lianwen(1); Liu, Weifeng(2); Li, Xuelong(3)Source: IEEE Transactions on Multimedia Volume: 15 Issue: 4 DOI: 10.1109/TMM.2013.2238909 Published: 2013Abstract:With the rapid development of the cloud computing and mobile service, users expect a better experience through multimedia computing, such as automatic or semi-automatic personal image and video organization and intelligent user interface. These functions heavily depend on the success of image understanding, and thus large-scale image annotation has received intensive attention in recent years. The collaboration between mobile and cloud opens a new avenue for image annotation, because the heavy computation can be transferred to the cloud for immediately responding user actions. In this paper, we present a scheme for image annotation on the cloud, which transmits mobile images compressed by Hamming compressed sensing to the cloud and conducts semantic annotation through a novel Hessian regularized support vector machine on the cloud. We carefully explained the rationality of Hessian regularization for encoding the local geometry of the compact support of the marginal distribution and proved that Hessian regularized support vector machine in the reproducing kernel Hilbert space is equivalent to conduct Hessian regularized support vector machine in the space spanned by the principal components of the kernel principal component analysis. We conducted experiments on the PASCAL VOC'07 dataset and demonstrated the effectiveness of Hessian regularized support vector machine for large-scale image annotation. ? 1999-2012 IEEE.Accession Number: 20132116361723 -
Record 47 of
Title:Image hiding scheme by use of rotating squared sub-image in the gyrator transform domains
Author(s):Liu, Zhengjun(1,2); Li, She(3); Liu, Wei(4); Liu, Wanyu(1); Liu, Shutian(4)Source: Optics and Laser Technology Volume: 45 Issue: 1 DOI: 10.1016/j.optlastec.2012.07.004 Published: February 2013Abstract:An image hiding scheme is proposed based on the scrambling process composed of the rotation of the squared sub-image in the gyrator transform domains. A squared sub-image is first selected from a secret image and is rotated along an axis, which is the center line or diagonal line of the sub-image. The rotation operation will be performed iteratively for utilizing more parameters being regarded as key. Subsequently, the rotated image is converted by gyrator transform into a complex function. The amplitude distribution and phase distribution are scrambled by the rotation process with different parameters. Finally, the encrypted image is obtained by gyrator transform from the rotated complex function. Numerical simulation is given to test the validity of the image encryption algorithm. ? 2012 Elsevier Ltd.Accession Number: 20123915470247 -
Record 48 of
Title:Temperature compensation in full optical fiber current transformer using signal processing
Author(s):Li, Yuanyuan(1); Yang, Xiaojun(1); Xu, Jintao(1); Wang, Yingli(1)Source: Proceedings - 6th International Symposium on Computational Intelligence and Design, ISCID 2013 Volume: 2 Issue: DOI: 10.1109/ISCID.2013.170 Published: 2013Abstract:Error caused by temperature change is one of the major reasons for restricting the promotion of Full Optical Fiber Current Transformer (FOCT) at present. This article analyzes how the change of temperature makes impact on the parameters of optical devices, such as the original phase of quarter-wave plate and the Verdet constant of the fiber. After the experimental data are sampled in a temperature range from +20°C to +70°C, the relational model of error and temperature can be set up by adopting least-square method. After compensation based on this model, error could be decreased, meanwhile the precision of FOCT could be improved. Indeed, the experimental results verify that the accuracy of FOCT can be improved four times, compared with the original output without compensation. This method of compensation is quite effective to reduce the temperature influences on the FOCT to improve the accuracy. ? 2013 IEEE.Accession Number: 20142217758445