Robust and Efficient Image Alignment Method Using the Student-t Distribution

Pixel-based image alignment has always struggled to simultaneously reject outliers, avoid local minima and run quickly. There are many robust cost functions that perform well in terms of rejecting outliers, but they can yield unstable results during long image sequences as a result of their inability to adjust to changes in image content. In this paper, we propose a parameterised student-t cost function that can interpolate between two cost functions that are amongst the most widely used ones in image alignment problems, the $L2$ norm (quadratic function) and the Cauchy-Lorentzian function. We also propose a parameter estimation method that helps to find optimal parameters for the proposed cost function for a video. Experiments prove that the proposed approach can estimate the alignment variable accurately relative to the existing cost functions without demanding a higher computational cost.

[1]  Thomas Sikora,et al.  Evaluation of pixel- and motion vector-based global motion estimation for camera motion characterization , 2009, 2009 10th Workshop on Image Analysis for Multimedia Interactive Services.

[2]  Yaakov Bar-Shalom,et al.  Expected likelihood for tracking in clutter with particle filters , 2002, SPIE Defense + Commercial Sensing.

[3]  David G. Lowe,et al.  Object recognition from local scale-invariant features , 1999, Proceedings of the Seventh IEEE International Conference on Computer Vision.

[4]  Jian Zheng,et al.  Robust Non-Rigid Point Set Registration Using Student's-t Mixture Model , 2014, PloS one.

[5]  Yuzhuo Zhong,et al.  A pre-analysis method for robust global motion estimation , 1999, Proceedings 1999 International Conference on Image Processing (Cat. 99CH36348).

[6]  Jan Flusser,et al.  Image registration methods: a survey , 2003, Image Vis. Comput..

[7]  J.-Y. Bouguet,et al.  Pyramidal implementation of the lucas kanade feature tracker , 1999 .

[8]  Steven Mills,et al.  Real-time aerial image mosaicing , 2010, 2010 25th International Conference of Image and Vision Computing New Zealand.

[9]  Angelika Foerster.  An R And S Plus Companion To Applied Regression , 2016 .

[10]  Simon Baker,et al.  Lucas-Kanade 20 Years On: A Unifying Framework , 2004, International Journal of Computer Vision.

[11]  Charles V. Stewart,et al.  Robust Parameter Estimation in Computer Vision , 1999, SIAM Rev..

[12]  Aristidis Likas,et al.  The mixtures of Student's t-distributions as a robust framework for rigid registration , 2009, Image Vis. Comput..

[13]  James R. Bergen,et al.  Visual odometry , 2004, Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2004. CVPR 2004..

[14]  Christopher Hunt,et al.  Notes on the OpenSURF Library , 2009 .

[15]  Michael J. Black,et al.  The Robust Estimation of Multiple Motions: Parametric and Piecewise-Smooth Flow Fields , 1996, Comput. Vis. Image Underst..

[16]  Takeo Kanade,et al.  An Iterative Image Registration Technique with an Application to Stereo Vision , 1981, IJCAI.

[17]  K. V. Arya,et al.  Image registration using robust M-estimators , 2007, Pattern Recognit. Lett..

[18]  Michael J. Black,et al.  A framework for the robust estimation of optical flow , 1993, 1993 (4th) International Conference on Computer Vision.

[19]  Filippo Vella,et al.  Digital image stabilization by adaptive block motion vectors filtering , 2002, IEEE Trans. Consumer Electron..

[20]  Chiou-Ting Hsu,et al.  Mosaics of video sequences with moving objects , 2004, Signal Process. Image Commun..

[21]  Robert T. Collins,et al.  An Open Source Tracking Testbed and Evaluation Web Site , 2005 .

[22]  Lap-Pui Chau,et al.  A Nonlinear $L _{1}$-Norm Approach for Joint Image Registration and Super-Resolution , 2009, IEEE Signal Processing Letters.

[23]  Frederick R. Forst,et al.  On robust estimation of the location parameter , 1980 .