Recovering shape and motion by a dynamic system for low-rank matrix approximation in L1 norm

To recover motion and shape matrices from a matrix of tracking feature points on a rigid object under orthography, we can do low-rank matrix approximation of the tracking matrix with its each column minus the row mean vector of the matrix. To obtain the row mean vector, usually 4-rank matrix approximation is used to recover the missing entries. Then, 3-rank matrix approximation is used to recover the shape and motion. Obviously, the procedure is not convenient. In this paper, we build a cost function which calculates the shape matrix, motion matrix as well as the row mean vector at the same time. The function is in L1 norm, and is not smooth everywhere. To optimize the function, a continuous-time dynamic system is newly proposed. With time going on, the product of the shape and rotation matrices becomes closer and closer, in L1-norm sense, to the tracking matrix with each its column minus the mean vector. A parameter is implanted into the system for improving the calculating efficiency, and the influence of the parameter on approximation accuracy and computational efficiency are theoretically studied and experimentally confirmed. The experimental results on a large number of synthetic data and a real application of structure from motion demonstrate the effectiveness and efficiency of the proposed method. The proposed system is also applicable to general low-rank matrix approximation in L1 norm, and this is also experimentally demonstrated.

[1]  Guanghui Wang,et al.  Stratification Approach for 3-D Euclidean Reconstruction of Nonrigid Objects From Uncalibrated Image Sequences , 2008, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics).

[2]  Tomás Pajdla,et al.  3D reconstruction by fitting low-rank matrices with missing data , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[3]  Takayuki Okatani,et al.  On the Wiberg Algorithm for Matrix Factorization in the Presence of Missing Components , 2007, International Journal of Computer Vision.

[4]  Yeung Sam Hung,et al.  Multi-stage 3D reconstruction under circular motion , 2007, Image Vis. Comput..

[5]  Peter Janssen,et al.  Extracting 3D structure from disparity , 2006, Trends in Neurosciences.

[6]  Yi Ma,et al.  The Augmented Lagrange Multiplier Method for Exact Recovery of Corrupted Low-Rank Matrices , 2010, Journal of structural biology.

[7]  Andrew W. Fitzgibbon,et al.  Damped Newton algorithms for matrix factorization with missing data , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[8]  Zhi-Hua Zhou,et al.  Generalized Low-Rank Approximations of Matrices Revisited , 2010, IEEE Transactions on Neural Networks.

[9]  Ventseslav Sainov,et al.  3-D Time-Varying Scene Capture Technologies—A Survey , 2007, IEEE Transactions on Circuits and Systems for Video Technology.

[10]  Nikos Paragios,et al.  Handbook of Mathematical Models in Computer Vision , 2005 .

[11]  Pei Chen,et al.  Heteroscedastic Low-Rank Matrix Approximation by the Wiberg Algorithm , 2008, IEEE Transactions on Signal Processing.

[12]  Shuzhi Sam Ge,et al.  $k$-NS: A Classifier by the Distance to the Nearest Subspace , 2011, IEEE Transactions on Neural Networks.

[13]  John Wright,et al.  Robust Principal Component Analysis: Exact Recovery of Corrupted Low-Rank Matrices via Convex Optimization , 2009, NIPS.

[14]  Takeo Kanade,et al.  Robust L/sub 1/ norm factorization in the presence of outliers and missing data by alternative convex programming , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[15]  S. Yun,et al.  An accelerated proximal gradient algorithm for nuclear norm regularized linear least squares problems , 2009 .

[16]  S. Zucker,et al.  Differential Geometry from the Frenet Point of View: Boundary Detection, Stereo, Texture and Color , 2006, Handbook of Mathematical Models in Computer Vision.

[17]  Anders P. Eriksson,et al.  Efficient computation of robust low-rank matrix approximations in the presence of missing data using the L1 norm , 2010, 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[18]  Aeron Buchanan Morgan,et al.  Investigation into Matrix Factorization when Elements are Unknown , 2004 .

[19]  Yiguang Liu,et al.  A novel and quick SVM-based multi-class classifier , 2006, Pattern Recognit..

[20]  Thomas Brox,et al.  Diffusion Filters and Wavelets: What Can They Learn from Each Other? , 2006, Handbook of Mathematical Models in Computer Vision.

[21]  M.T. El-Melegy,et al.  A variational technique for 3D reconstruction from multiple views , 2007, 2007 International Conference on Computer Engineering & Systems.

[22]  Takeo Kanade,et al.  Shape and motion from image streams under orthography: a factorization method , 1992, International Journal of Computer Vision.

[23]  Takeo Kanade,et al.  A sequential factorization method for recovering shape and motion from image streams , 1997, IEEE Trans. Pattern Anal. Mach. Intell..

[24]  Gang Hua,et al.  A Hierarchical Visual Model for Video Object Summarization , 2010, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[25]  Emmanuel J. Candès,et al.  A Singular Value Thresholding Algorithm for Matrix Completion , 2008, SIAM J. Optim..

[26]  Yiguang Liu,et al.  A functional neural network computing some eigenvalues and eigenvectors of a special real matrix , 2005, Neural Networks.

[27]  Jieping Ye,et al.  Generalized Low Rank Approximations of Matrices , 2004, Machine Learning.

[28]  Michael J. Black,et al.  Robust Principal Component Analysis for Computer Vision , 2001, ICCV.

[29]  Aaron Hertzmann,et al.  Nonrigid Structure-from-Motion: Estimating Shape and Motion with Hierarchical Priors , 2008, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[30]  John Wright,et al.  RASL: Robust Alignment by Sparse and Low-Rank Decomposition for Linearly Correlated Images , 2012, IEEE Trans. Pattern Anal. Mach. Intell..

[31]  Pei Chen,et al.  Optimization Algorithms on Subspaces: Revisiting Missing Data Problem in Low-Rank Matrix , 2008, International Journal of Computer Vision.

[32]  Derong Liu.  Editorial: the IEEE transactions on neural networks 2010 and beyond , 2010, IEEE Trans. Neural Networks.

[33]  G. Sapiro,et al.  A collaborative framework for 3D alignment and classification of heterogeneous subvolumes in cryo-electron tomography. , 2013, Journal of structural biology.

[34]  Yiguang Liu,et al.  A concise functional neural network computing the largest modulus eigenvalues and their corresponding eigenvectors of a real skew matrix , 2006, Theor. Comput. Sci..

[35]  Andrea Montanari,et al.  Matrix completion from a few entries , 2009, ISIT.

[36]  Yonghuai Liu,et al.  Automatic Range Image Registration in the Markov Chain , 2010, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[37]  Stan Z. Li.  Markov Random Field Modeling in Image Analysis , 2009, Advances in Pattern Recognition.

[38]  R. Hartley,et al.  PowerFactorization : 3D reconstruction with missing or uncertain data , 2003 .