A New Approach to Probabilistic Image Modeling with Multidimensional Hidden Markov Models

This paper presents a novel multi-dimensional hidden Markov model approach to tackle the complex issue of image modeling. We propose a set of efficient algorithms that avoids the exponential complexity of regular multidimensional HMMs for the most frequent algorithms (Baum-Welch and Viterbi) due to the use of a random dependency tree (DT-HMM). We provide the theoretical basis for these algorithms, and we show that their complexity remains as small as in the uni-dimensional case. A number of possible applications are given to illustrate the genericity of the approach. Experimental results are also presented in order to demonstrate the potential of the proposed DTHMM for common image analysis tasks such as object segmentation, and tracking.

[1]  Kenneth Rose,et al.  Deformable face mapping for person identification , 2003, Proceedings 2003 International Conference on Image Processing (Cat. No.03CH37429).

[2]  Joakim Jiten,et al.  Probabilistic image modeling with dependency-tree hidden Markov models , 2006 .

[3]  Lawrence R. Rabiner,et al.  A tutorial on hidden Markov models and selected applications in speech recognition , 1989, Proc. IEEE.

[4]  Pietro Perona,et al.  Recognition by Probabilistic Hypothesis Construction , 2004, ECCV.

[5]  Frederick Jelinek,et al.  Basic Methods of Probabilistic Context Free Grammars , 1992 .

[6]  P. Gader,et al.  Generalized Hidden Markov Models — Part I : Theoretical Frameworks , 2008 .

[7]  Daniel P. Huttenlocher,et al.  Image segmentation using local variation , 1998, Proceedings. 1998 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No.98CB36231).

[8]  Stéphane Marchand-Maillet,et al.  Approximate Viterbi decoding for 2D-hidden Markov models , 2000, 2000 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.00CH37100).

[9]  Jiří Matas,et al.  Computer Vision - ECCV 2004 , 2004, Lecture Notes in Computer Science.

[10]  L. Baum,et al.  Statistical Inference for Probabilistic Functions of Finite State Markov Chains , 1966 .

[11]  Frederick Jelinek,et al.  Statistical methods for speech recognition , 1997 .

[12]  Robert M. Gray,et al.  Image classification by a two dimensional hidden Markov model , 1999, 1999 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings. ICASSP99 (Cat. No.99CH36258).

[13]  Paul D. Gader,et al.  Generalized hidden Markov models. I. Theoretical frameworks , 2000, IEEE Trans. Fuzzy Syst..

[14]  Yoram Singer,et al.  The Hierarchical Hidden Markov Model: Analysis and Applications , 1998, Machine Learning.

[15]  J. Baker.  Trainable grammars for speech recognition , 1979 .

[16]  Alex Pentland,et al.  Coupled hidden Markov models for complex action recognition , 1997, Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[17]  Laveen N. Kanal,et al.  Markov mesh models , 1980 .

[18]  Joo-Hwee Lim,et al.  Semantics Discovery for Image Indexing , 2004, ECCV.

[19]  Roberto Pieraccini,et al.  Dynamic planar warping for optical character recognition , 1992, [Proceedings] ICASSP-92: 1992 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[20]  Patrick Bouthemy,et al.  Extraction of Semantic Dynamic Content from Videos with Probabilistic Motion Models , 2004, ECCV.

[21]  Tom E. Bishop,et al.  Blind Image Restoration Using a Block-Stationary Signal Model , 2006, 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings.