-
Record 169 of
Title:Joint embedding learning and sparse regression: A framework for unsupervised feature selection
Author(s):Hou, Chenping(1); Nie, Feiping(2); Li, Xuelong(3); Yi, Dongyun(1); Wu, Yi(1)
Source: IEEE Transactions on Cybernetics Volume: 44 Issue: 6 DOI: 10.1109/TCYB.2013.2272642 Published: June 2014
Abstract:Feature selection has aroused considerable research interests during the last few decades. Traditional learning-based feature selection methods separate embedding learning and feature ranking. In this paper, we propose a novel unsupervised feature selection framework, termed as the joint embedding learning and sparse regression (JELSR), in which the embedding learning and sparse regression are jointly performed. Specifically, the proposed JELSR joins embedding learning with sparse regression to perform feature selection. To show the effectiveness of the proposed framework, we also provide a method using the weight via local linear approximation and adding the 2,1-norm regularization, and design an effective algorithm to solve the corresponding optimization problem. Furthermore, we also conduct some insightful discussion on the proposed feature selection approach, including the convergence analysis, computational complexity, and parameter determination. In all, the proposed framework not only provides a new perspective to view traditional methods but also evokes some other deep researches for feature selection. Compared with traditional unsupervised feature selection methods, our approach could integrate the merits of embedding learning and sparse regression. Promising experimental results on different kinds of data sets, including image, voice data and biological data, have validated the effectiveness of our proposed algorithm. ? 2013 IEEE.
Accession Number: 20142217766266
-
Record 170 of
Title:Research on measurement and correction of a fish-eye image distortion
Author(s):Wang, Zefeng(1); Lei, Yangjie(1); Zhang, Zhi(1); Zhang, Zhaohui(1); Zhang, Hui(1); Huang, Jijiang(1); Yi, Bo(1); Liao, Jiawen(1)
Source: Proceedings of SPIE - The International Society for Optical Engineering Volume: 9282 Issue: DOI: 10.1117/12.2068149 Published: 2014
Abstract:Fisheye lenses have the advantages of short focal length and large field of view. However, by using the "non-similar" imaging principle, they artificially introduce a large barrel distortion. In order to improve the quality of the images correction of distortion is required. This article analyzes the polar distortion correction model, raised a simple distortion coefficient calibration method and the use of bilinear interpolation method for gray level interpolation. Compared to other methods, this method is easier to reinforce and achieves high accuracy, and it can be easily implemented in the hardware system. At the end of the paper we introduced a device correction for a fisheye CCD camera. Based on the original data, a distortion correction model is established. In order to minimize the error, the correction was divided into three sections, and the image is well recovered. ? 2014 SPIE.
Accession Number: 20150800543906
-
Record 171 of
Title:Re-texturing by intrinsic video
Author(s):Shen, Jianbing(1); Yan, Xing(1); Chen, Lin(1); Sun, Hanqiu(2); Li, Xuelong(3)
Source: Information Sciences Volume: 281 Issue: DOI: 10.1016/j.ins.2014.02.134 Published: October 10, 2014
Abstract:In this paper, we present a novel re-texturing approach using intrinsic video. Our approach first indicates the regions of interest by contour-aware layer segmentation. The intrinsic video including reflectance and illumination components within the segmented region is recovered by our weighted energy optimization. We then compute the texture coordinates in key frames and the normals for the re-textured region using the optimization approach we develop. Meanwhile, the texture coordinates in non-key frames are optimized by our energy function. When the target sample texture is specified, the re-textured video is finally created by multiplying the re-textured reflectance component with the original illumination component within the replaced region. As shown in our experimental results, our method can produce high quality video re-texturing results with a variety of sample textures, and also the lighting and shading effects of the original videos are well preserved after re-texturing. ? 2014 Elsevier Inc. All rights reserved.
Accession Number: 20143117996579
-
Record 172 of
Title:Design of unobscured three-mirror optical system by applying vector wavefront aberration theory
Author(s):Zou, Gangyi(1); Fan, Xuewu(1); Pang, Zhihai(1); Feng, Liangjie(1); Ren, Guorui(1)
Source: Hongwai yu Jiguang Gongcheng/Infrared and Laser Engineering Volume: 43 Issue: 2 DOI: Published: February 2014
Abstract:The traditional unobscured three-mirror optical system is an intrinsically rotationally symmetric optical system with an offset aperture stop, a biased input field, or both of them, so off-axis sections of rotationally symmetric aspheric parent surface are ineluctable. Using the conclusion of vector wavefront aberration theory, a new unobscured three-mirror system by tilted the rotationally symmetric aspheric mirror was presented. The design reason and step of this system was analyzed, and then a system with effective focal length of 1 000 mm, field of view of 10° ×20° and F -number 10 was designed. The volume of system (Length×Wide×Height) less than 350 mm×350 mm×120 mm and image qualities of the example are near diffraction limit. Compared with other unobscured three-mirror system, the most prominent advantage of this system is that using tilted rotationally symmetric aspheric mirror to achieve unobscured style, thus reducing cost of the system.
Accession Number: 20141317523540
-
Record 173 of
Title:Improvement of image deblurring for opto-electronic joint transform correlator under projective motion vector estimation
Author(s):Xiao, Xiao(1); Zhao, Hui(2); Zhang, Yang(1)
Source: Optics Communications Volume: 321 Issue: DOI: 10.1016/j.optcom.2014.02.006 Published: June 15, 2014
Abstract:In this paper we propose an efficient algorithm to improve the performance of image deblurring based on opto-electronic joint transform correlator (JTC) that is capable of detecting the motion vector of a space camera. Firstly, the motion vector obtained from JTC is divided into many sub-motion vectors according to the projective motion path, which represents the degraded image as an integration of the clear scene under a sequence of planar projective transforms. Secondly, these sub-motion vectors are incorporated into the projective motion Richardson-Lucy (RL) algorithm to improve deblurred results. The simulation results demonstrate the effectiveness of the algorithm and the influence of noise on the algorithm performance is also statically analyzed. ? 2014 Elsevier B.V.
Accession Number: 20141017428751
-
Record 174 of
Title:Learning deep and wide: A spectral method for learning deep networks
Author(s):Shao, Ling(1,2); Wu, Di(2); Li, Xuelong(3)
Source: IEEE Transactions on Neural Networks and Learning Systems Volume: 25 Issue: 12 DOI: 10.1109/TNNLS.2014.2308519 Published: December 1, 2014
Abstract:Building intelligent systems that are capable of extracting high-level representations from high-dimensional sensory data lies at the core of solving many computer vision-related tasks. We propose the multispectral neural networks (MSNN) to learn features from multicolumn deep neural networks and embed the penultimate hierarchical discriminative manifolds into a compact representation. The low-dimensional embedding explores the complementary property of different views wherein the distribution of each view is sufficiently smooth and hence achieves robustness, given few labeled training data. Our experiments show that spectrally embedding several deep neural networks can explore the optimum output from the multicolumn networks and consistently decrease the error rate compared with a single deep network. ? 2012 IEEE.
Accession Number: 20144900289124
-
Record 175 of
Title:Refraction angle extracting strategy for fan-beam differential phase contrast CT
Author(s):Ye, Renzhen(1); Tang, Yi(2); Lu, Xiaoqiang(3)
Source: Neurocomputing Volume: 141 Issue: DOI: 10.1016/j.neucom.2014.03.040 Published: October 2, 2014
Abstract:In this paper, the fan-beam differential phase contrast computed tomography (DPC-CT) reconstruction method is studied. We first present a new vision of how to implement the Reverse-Projection (RP) method to extract the refraction-angle data efficiently in fan-beam geometry, and then provide a Katsevich-type formula for fan-beam DPC-CT reconstruction. The proposed method has two key properties. First, it is essentially a filtered back projection (FBP) reconstruction formula. Second, it can deal with incomplete data sets. The main contributions of this paper lie in the following three aspects: First, the physical principle of the bent-grating based fan-beam DPC imaging is discussed and the RP-method is extended to the fan-beam case. Second, an implementation strategy of Katsevich algorithm for fan-beam DPC-CT is proposed. Third, a semi-quantitative research on the influence of the approximation errors introduced by the RP-method is carried out by using several numerical simulations. It should be pointed out that the RP-method will certainly introduce some errors. The effect of these errors on our reconstruction algorithm is discussed by several numerical simulations. ? 2014 Elsevier B.V.
Accession Number: 20142317789260
-
Record 176 of
Title:Efficient dictionary learning for visual categorization
Author(s):Tang, Jun(1); Shao, Ling(2); Li, Xuelong(3)
Source: Computer Vision and Image Understanding Volume: 124 Issue: DOI: 10.1016/j.cviu.2014.02.007 Published: July 2014
Abstract:We propose an efficient method to learn a compact and discriminative dictionary for visual categorization, in which the dictionary learning is formulated as a problem of graph partition. Firstly, an approximate kNN graph is efficiently computed on the data set using a divide-and-conquer strategy. And then the dictionary learning is achieved by seeking a graph topology on the resulting kNN graph that maximizes a submodular objective function. Due to the property of diminishing return and monotonicity of the defined objective function, it can be solved by means of a fast greedy-based optimization. By combing these two efficient ingredients, we finally obtain a genuinely fast algorithm for dictionary learning, which is promising for large-scale datasets. Experimental results demonstrate its encouraging performance over several recently proposed dictionary learning methods. ? 2014 Elsevier Inc. All rights reserved.
Accession Number: 20142517827024
-
Record 177 of
Title:Action recognition by spatio-temporal oriented energies
Author(s):Zhen, Xiantong(1,2); Shao, Ling(1,2); Li, Xuelong(3)
Source: Information Sciences Volume: 281 Issue: DOI: 10.1016/j.ins.2014.05.021 Published: October 10, 2014
Abstract:In this paper, we present a unified representation based on the spatio-temporal steerable pyramid (STSP) for the holistic representation of human actions. A video sequence is viewed as a spatio-temporal volume preserving all the appearance and motion information of an action in it. By decomposing the spatio-temporal volumes into band-passed sub-volumes, the spatio-temporal Laplacian pyramid provides an effective technique for multi-scale analysis of video sequences, and spatio-temporal patterns with different scales could be well localized and captured. To efficiently explore the underlying local spatio-temporal orientation structures at multiple scales, a bank of three-dimensional separable steerable filters are conducted on each of the sub-volume from the Laplacian pyramid. The outputs of the quadrature pair of steerable filters are squared and summed to yield a more robust oriented energy representation. To be further invariant and compact, a spatio-temporal max pooling operation is performed between responses of the filtering at adjacent scales and over spatio-temporal neighbourhoods. In order to capture the appearance, local geometric structure and motion of an action, we apply the STSP on the intensity, 3D gradients and optical flow of video sequences, yielding a unified holistic representation of human actions. Taking advantage of multi-scale, multi-orientation analysis and feature pooling, STSP produces a compact but informative and invariant representation of human actions. We conduct extensive experiments on the KTH, UCF Sports and HMDB51 datasets, which shows the unified STSP achieves comparable results with the state-of-the-art methods. ? 2014 Elsevier Inc. All rights reserved.
Accession Number: 20143117996602
-
Record 178 of
Title:Efficient dictionary learning for visual categorization
Author(s):Tang, Jun(1); Shao, Ling(2); Li, Xuelong(3)
Source: Computer Vision and Image Understanding Volume: 124 Issue: DOI: 10.1016/j.cviu.2014.02.007 Published: July 2014
Abstract:We propose an efficient method to learn a compact and discriminative dictionary for visual categorization, in which the dictionary learning is formulated as a problem of graph partition. Firstly, an approximate kNN graph is efficiently computed on the data set using a divide-and-conquer strategy. And then the dictionary learning is achieved by seeking a graph topology on the resulting kNN graph that maximizes a submodular objective function. Due to the property of diminishing return and monotonicity of the defined objective function, it can be solved by means of a fast greedy-based optimization. By combing these two efficient ingredients, we finally obtain a genuinely fast algorithm for dictionary learning, which is promising for large-scale datasets. Experimental results demonstrate its encouraging performance over several recently proposed dictionary learning methods. ? 2014 Elsevier Inc. All rights reserved.
Accession Number: 20142417815389
-
Record 179 of
Title:Ego motion guided particle filter for vehicle tracking in airborne videos
Author(s):Cao, Xianbin(1); Gao, Changcheng(1); Lan, Jinhe(2); Yuan, Yuan(3); Yan, Pingkun(3)
Source: Neurocomputing Volume: 124 Issue: DOI: 10.1016/j.neucom.2013.07.014 Published: January 26, 2014
Abstract:Tracking in airborne circumstances is receiving more and more attention from researchers, and it has become one of the most important components in video surveillance for its advantage of better mobility, larger surveillance scope and so on. However, airborne vehicle tracking is very challenging due to the factors such as platform motion, scene complexity, etc. In this paper, to address these problems, a new framework based on Kanade-Lucas-Tomasi (KLT) features and particle filter is proposed. KLT features are tracked throughout the video sequence. At the beginning of video tracking, a strategy based on motion consistence with RANSAC is utilized to separate background KLT features. The grouping of background features helps estimate the ego motion of the platform and the estimation is then incorporated into the prediction step in particle filter. Color similarity and Hu moments are used in the measurement model to assign the weights of particles. Our experimental results demonstrated that the proposed method outperformed the other tracking methods. ? 2013 Elsevier B.V.
Accession Number: 20134316889887
-
Record 180 of
Title:Fabrication and annealing optimization of oxygen-implanted Yb 3+-doped phosphate glass planar waveguides
Author(s):Liu, Chun-Xiao(1,2); Xu, Jun(3); Li, Wei-Nan(2); Xu, Xiao-Li(1); Guo, Hai-Tao(2); Wei, Wei(2,4); Wu, Gen-Gen(1); Hu, Yue(1); Peng, Bo(2,4)
Source: Optics and Laser Technology Volume: 63 Issue: DOI: 10.1016/j.optlastec.2014.03.014 Published: November 2014
Abstract:Optical planar waveguides in Yb3+-doped phosphate glasses are fabricated by (5.0+6.0) MeV O3+ ion implantation at fluences of (4.0+8.0)×1014 ions/cm2. The annealing treatment is carried out to optimize waveguide performances. The prism-coupling and end-face coupling methods are used to measure the dark-mode spectra and near-field intensity distributions before and after annealing at 350 °C for 60 min, respectively. The refractive index profile of the planar waveguide is obtained based on the reflectivity calculation method. The micro-Raman spectrum of the waveguide is in agreement with that of the bulk, exhibiting possible applications for integrated active photonic devices. ? 2014 Elsevier Ltd.
Accession Number: 20141717604259
