Towards a Coherent Statistical Framework for Dense Deformable Template Estimation

Stéphanie Allassonnière1, Yali Amit2, Alain Trouvé3
1Université Paris XIII, France
2University of Chicago, USA,,
3USAEcole Normale Supérieur , Cachan, and Université Paris XIII , France

Tóm tắt

SummaryThe problem of estimating probabilistic deformable template models in the field of computer vision or of probabilistic atlases in the field of computational anatomy has not yet received a coherent statistical formulation and remains a challenge. We provide a careful definition and analysis of a well-defined statistical model based on dense deformable templates for grey level images of deformable objects. We propose a rigorous Bayesian framework for which we prove asymptotic consistency of the maximum a posteriori estimate and which leads to an effective iterative estimation algorithm of the geometric and photometric parameters in the small sample setting. The model is extended to mixtures of finite numbers of such components leading to a fine description of the photometric and geometric variations of an object class. We illustrate some of the ideas with images of handwritten digits and apply the estimated models to classification through maximum likelihood.

Từ khóa


Tài liệu tham khảo

Amit, 2002, 2d Object Detection and Recognition: Models, Algorithms and Networks, 10.7551/mitpress/1006.001.0001

Amit, 2004, A coarse-to-fine strategy for multi-class shape detection, IEEE Trans. Pattn Anal. Mach. Intell., 26, 1606, 10.1109/TPAMI.2004.111

Amit, 1991, Structural image restoration through deformable template, J. Am. Statist. Ass., 86, 376, 10.1080/01621459.1991.10475053

Amit, 2004, Pop: patchwork of parts models for object recognition

Amit, 2005, Generative models for labeling multi-object configurations in images

Beg, 2005, Computing large deformation metric mappings via geodesic flows of diffeomorphisms, Int. J. Comput. Vis., 61, 139, 10.1023/B:VISI.0000043755.93987.aa

Cootes, 1998, Proc. 5th Eur. Conf. Computer Vision, Berlin, 484

Dempster, 1977, Maximum likelihood from incomplete data via the EM algorithm, J. R. Statist. Soc. B, 38, 1

Fergus, 2003, Proc. Computer Vision and Pattern Recognition Conf, 264

Glasbey, 2001, A penalised likelihood approach to image warping, J. R. Statist. Soc. B, 63, 465, 10.1111/1467-9868.00295

Grenander, 1998, Computational anatomy: an emerging discipline, Q. Appl. Math., 56, 617, 10.1090/qam/1668732

Hastie, 1998, Metrics and models for handwritten character recognition, Statist. Sci., 13, 54, 10.1214/ss/1028905973

Hastie, 2001, The Elements of Statistical Learning Theory, 10.1007/978-0-387-21606-5

Jennison, 2001, Discussion on ‘A penalized likelihood approach to image warping’ (by C. A. Glasbey and K. V. Mardia), J. R. Statist. Soc. B, 63, 493

Joshi, 2004, Unbiased diffeomorphic atlas construction for computational anatomy, Neuroimage, 23, S151, 10.1016/j.neuroimage.2004.07.068

Klassen, 2004, Analysis of planar shapes using geodesic paths on shape spaces, IEEE Trans. Pattn Anal. Mach. Intell., 26, 372, 10.1109/TPAMI.2004.1262333

Li, 2003, Proc. Int. Conf. Computer Vision, 1134

Marsland, 2004, Constructing diffeomorphic representations for the groupwise analysis of non-rigid registrations of medical images, IEEE Trans. Med. Imgng, 23

Michor, 2006, Riemannian geometries on spaces of planar curves, J. Eur. Math. Soc., 8, 1, 10.4171/jems/37

Miller, 2002, On the metrics and Euler-Lagrange equations of computational anatomy, A. Rev. Biomed. Engng, 4, 375, 10.1146/annurev.bioeng.4.092101.125733

Neal, 1998, Learning in Graphical Models, 355, 10.1007/978-94-011-5014-9_12

Trouvé, 2005, Metamorphoses through lie group action, Found. Computnl Math., 5, 173, 10.1007/s10208-004-0128-z

Van Der Vaart, 1998, Asymptotic Statistics, 10.1017/CBO9780511802256