Abstract
The following problem arises in computer vision, diagnostic medical imaging and remote sensing: At each pixel in an image a vector of observations is measured. The distribution of these measurements is modeled by a mixture of certain pure class distributions and the goal is to estimate the mixing proportions of the classes by pixel in the image together with any unknown parameters in the pure class distributions. In many problems of this type it is appropriate to incorporate constraints on the mixing proportions. This paper deals with spatial smoothness constraints. An estimation methodology using penalized likelihood with multiple smoothing parameters is applied. Numerical methods for evaluating parameters in the model are developed. We make use of the Expectation Maximization (EM) algorithm. An importance sampling technique for approximating the effective degrees of freedom of the model is also described. The methodology is illustrated with some examples.
The research was supported in part by the Dept. of Energy under FG0685-ER2500 by the National Cancer Institute under 2P01-CA-42045. I am grateful to David Haynor for some valuable discussions.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Smith M.O. Adams J.B. and Johnson P. Spectral mixture modelling, a new analysis of rock and soil types at the viking lander 1 site. J. Geophys. Res., 91:8098–8112, 1986.
Dempster A.P. and Laird N.M. ad Rubin D.B. Maximum likelihood from incomplete data via the em algorithm. J. Royal Stat. Soc. Ser. B, 39:1–38, 1977.
Gu C. and Wahba G. Minimizing gcv or gml scores with multiple smoothing parameters via the newton method. Technical Report 847, Department of Statistics, University of Wisconsin, Madison, 1988.
Titterington D.M., Smith A.F.M., and Makov U.E. Statistical Analysis of Finite Mixture Distributions. John Wiley and Sons, Inc., 1985.
Isaacson E. and Keller H.B. Analysis of Numerical Methods. J. Wiley and Sons, New York, 1966.
O’Sullivan F. Two-dimensional laplacian smoothing by iterative methods. J. Amer. Statist. Assoc, (in press), 1990.
Golub G.H. and Van Loan C.F. Matrix Computations. North Oxford Academic, New York, 1983.
Choi H.S., Haynor D.R., and Kim Y. Multivariate tissue classification of mri images for 3-d volume reconstruction — a statistical approach. SPIE Medical Imaging III: Image Processing, 1092:183–193, 1989.
Haslett J. Maximum likelihood discriminant analysis on the plane using a markovian model of spatial context. Pattern Recognition, 18:287–297, 1985.
Kent J.T. and Mardia K.V. Spatial classification using fuzzy membership models. IEEE Trans. Pattern Anal, and Machine Intell, 10:657–671, 1988.
Takeuchi K. Distribution of information statistics and a criterion for goodness of models. Mathematical Sciences, 153:12–18, 1976.
F. O’Sullivan. Discrete laplacian smoothing by fourier methods. J. Amer. Statist. Assoc, 1989. To appear.
F. O’Sullivan. Mixture estimation from multichannel image data. Technical Report 104, Department of Biostatistics and Statistics, University of Washington, 1989.
Redner R.A. and Walker H.F. Mixture densities, maximum likelihood, and the em algorithm. SIAM Review, 26:195–239, 1984.
Gancarz A.J. Reid M.J. and Albee A.L. Constrained least-squares analysis of petrologic problems with an application to lunar sample 12040. Earth and Planetary Science Letters, 17:433–445, 1973.
Vardi Y., Shepp L.A., and Kaufman L. A statistical model for positron emission tomography. J. Amer. Statist. Assoc, 80(389), 1985.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1992 Springer-Verlag New York, Inc.
About this paper
Cite this paper
O’Sullivan, F. (1992). Smooth Mixture Estimation from Multichannel Image Data. In: Page, C., LePage, R. (eds) Computing Science and Statistics. Springer, New York, NY. https://doi.org/10.1007/978-1-4612-2856-1_18
Download citation
DOI: https://doi.org/10.1007/978-1-4612-2856-1_18
Publisher Name: Springer, New York, NY
Print ISBN: 978-0-387-97719-5
Online ISBN: 978-1-4612-2856-1
eBook Packages: Springer Book Archive