Bayesian formulation of image patch matching using cross-correlation

A classical solution for matching two image patches is to use the cross-correlation coefficient. This works well if there is a lot of structure within the patches, but not so well if the patches are close to uniform. This means that some patches are matched with more confidence than others. By estimating this uncertainty more weight can be put on the confident matches than those that are more uncertain. In this paper we present a system that can learn the distribution of the correlation coefficient from a video sequence of an empty scene. No manual annotation of the video is needed. Two distributions functions are learned for two different cases: i) the correlation between an estimated background image and the current frame showing that background and ii) the correlation between an estimated background image and an unrelated patch.

[1]  Kentaro Toyama,et al.  Wallflower: principles and practice of background maintenance , 1999, Proceedings of the Seventh IEEE International Conference on Computer Vision.

[2]  Alex Pentland,et al.  Pfinder: Real-Time Tracking of the Human Body , 1997, IEEE Trans. Pattern Anal. Mach. Intell..

[3]  Rikard Berthilsson,et al.  Real Time Viterbi Optimization of Hidden Markov Models for Multi Target Tracking , 2007, 2007 IEEE Workshop on Motion and Video Computing (WMVC'07).

[4]  Olivier Bernier,et al.  Real Time Illumination Invariant Background Subtraction Using Local Kernel Histograms , 2006, BMVC.

[5]  P. Wayne Power,et al.  Understanding Background Mixture Models for Foreground Segmentation , 2002 .

[6]  Roberto Brunelli,et al.  Face Recognition: Features Versus Templates , 1993, IEEE Trans. Pattern Anal. Mach. Intell..

[7]  Stuart J. Russell,et al.  Image Segmentation in Video Sequences: A Probabilistic Approach , 1997, UAI.

[8]  Richard O. Duda,et al.  Pattern classification and scene analysis , 1974, A Wiley-Interscience publication.

[9]  W. Eric L. Grimson,et al.  Adaptive background mixture models for real-time tracking , 1999, Proceedings. 1999 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No PR00149).

[10]  J. C. Eilbeck.  Table errata: Higher transcendental functions. Vol. I, II (McGraw-Hill, New York, 1953) by A. Erdélyi, W. Magnus, F. Oberhettinger and F. G. Tricomi , 1971 .

[11]  P. Kohli,et al.  Efficiently solving dynamic Markov random fields using graph cuts , 2005, Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1.

[12]  Trevor Darrell,et al.  Background estimation and removal based on range and color , 1999, Proceedings. 1999 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No PR00149).

[13]  Michael Isard,et al.  Bayesian Object Localisation in Images , 2001, International Journal of Computer Vision.

[14]  N. L. Johnson,et al.  Continuous Univariate Distributions. , 1995 .

[15]  Yongtian Wang,et al.  An integrated background model for video surveillance based on primal sketch and 3D scene geometry , 2008, 2008 IEEE Conference on Computer Vision and Pattern Recognition.

[16]  Shun'ichi Kaneko,et al.  Robust image registration by increment sign correlation , 2002, Pattern Recognit..

[17]  Eva Möller,et al.  Statistical Properties and Control Algorithms of Recursive Quantile Estimators , 2000 .

[18]  H. Hotelling.  New Light on the Correlation Coefficient and its Transforms , 1953 .

[19]  Håkan Ardö.  Multi-target Tracking Using on-line Viterbi Optimisation and Stochastic Modelling , 2009 .

[20]  Milton Abramowitz,et al.  Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables , 1964 .

[21]  Nikos Paragios,et al.  Motion-based background subtraction using adaptive kernel density estimation , 2004, CVPR 2004.