On Optimal Filtering of Morphologically Smooth Discrete Random Sets and Related Open Problems
- 360 Downloads
Recently, it has been shown that morphological openings and closings can be viewed as consistent MAP estimators of (morphologically) smooth random sets immersed in clutter, or suffering from random dropouts —. These results hold for one-sided (union or intersection) noise. In the case of two-sided (union and intersection) noise we now know that this is no longer the case: in this more general setting, as it turns out, optimal estimators are not increasing operators . Then, how may one efficiently compute an optimal random set estimate in this more general setting? For 1-D and certain restricted 2-D random set models the answer is provided by the Viterbi algorithm . In the general case of 2-D random set models in two-sided noise the answer is unknown, and the task of finding it constitutes a challenging research problem.
Key wordsRandom Sets Morphological Filtering Opening Closing Dynamic Programming Viterbi Algorithm
Unable to display preview. Download preview PDF.
- D.G. KendallFoundations of a theory of random setsin Stochastic Geometry, E.F. Harding and D.G. Kendall, Eds., pp. 322–376. John Wiley, London, England, 1974.Google Scholar
- G. MatheronElements pour une theorie des Milieux PoreuxMasson, 1967.Google Scholar
- Gian-Carlo RotaOn the Foundations of Combinatorial Theory: I. Theory of Moebius FunctionsZ. Wahrscheinlichkeitstheorie und Verw. Gebiete, vol. 2, pp. 340–368, 1964. Also appears in Classic Papers in Combinatorics, I. Gessel and G-C. Rota (Eds.), Birkhauser, Boston,1987.MathSciNetzbMATHCrossRefGoogle Scholar
- H. Mathis ThomaBelief function computationsin Conditional Logic in Expert Systems, I.R. Goodman, M.M Gupta, H.T. Nguyen, and G.S. Rogers, Eds. 1991, pp. 269–308, Elsevier Science Publishers B.V. (North Holland).Google Scholar
- E.R. DoughertyOptimal Mean Square N-Observation Digital Morphological Filters - I. Optimal Binary FiltersComputer Vision, Graphics, and Image Processing: Image Understanding, vol. 55, pp. 36–54, 1992.Google Scholar
- R.M. Haralick, E.R. Dougherty, and P.L. KatzModel-based morphologyin Proc. SPIE, Vol. 1472, Orlando, Florida. Society for Optical Engineering, April 1991.Google Scholar
- E.R. Dougherty, A. Mathew, and V. SwarnakarA conditional-expectationbased implementation of the optimal mean-square binary morphological filterin Proc. SPIE, Vol. 1451, San Jose, California. Society for Optical Engineering, February 1991.Google Scholar
- E.R. Dougherty, ed.Mathematical Morphology in Image ProcessingMarcel Dekker, New York, 1993.Google Scholar
- A. Blake and A. ZissermanVisual ReconstructionMIT Press, Cambridge, Mass., 1987.Google Scholar