## Abstract

The problem of finding appropriate mathematical objects to model images is considered. Using the notion of a*completed graph* of a bounded function, which is a closed and bounded point set in the three-dimensional Euclidean space*R* _{3}, and exploring the*Hausdorff distance* between these point sets, a metric space*IM* _{D} of functions is defined. The main purpose is to show that the functions*f∈IM* _{D}, defined on the square*D*=[0,1]^{2}, are appropriate mathematical models of real world images.

The properties of the metric space*IM* _{D} are studied and methods of approximation for the purpose of image compression are presented.

The metric space*IM* _{D} contains the so-called*pixel functions* which are produced through digitizing images. It is proved that every function*f∈IM* _{D} may be digitized and represented by a pixel function*p* _{n}, with*n* pixels, in such a way that the distance between*f* and*p* _{n} is no greater than 2*n* ^{−1/2}.

It is advocated that the Hausdorff distance is the most natural one to measure the difference between two pixel representations of a given image. This gives a natural mathematical measure of the quality of the compression produced through different methods.

## AMS classification

Primary 41A46, 68U10 Secondary 41A99, 54D35## Key words and phrases

Hausdorff distance Hausdorff continuity Completed graph Pixel function Image compression Fractal transform operator## Preview

Unable to display preview. Download preview PDF.

## References

- 1.M. F. Barnsley (1988): Fractals Everywhere. Boston: Academic Press.Google Scholar
- 2.M. F. Barnsley, L. P. Hurd (1993): Fractal Image Compression. Wellesley, MA: Peters.Google Scholar
- 3.T. P. Boyanov, V. A. Popov (1970):
*On the widths of the space of continuous functions in the metric of Hausdorff*. Godishnik Sofia Univ. Mat. Fak.**63**:167–185 (in Bulgarian).MATHGoogle Scholar - 4.P. M. Diaconis, M. Shashahani (1986): Products of Random Matrices and Computer Image Generation. Contemporary Mathematics, vol. 50. Providence, RI: American Mathematical Society, pp. 173–182.Google Scholar
- 5.F. Hausdorff (1927). Mengenlehre. Berlin: Gruyter.Google Scholar
- 6.J. Hutchinson (1981):
*Fractals and self-similarity*. Indiana Univ. Math. J.,**30**:713–747.MATHMathSciNetCrossRefGoogle Scholar - 7.T. Kaiser (1978):
*A limit theorem for Markov chains in compact metric spaces with applications on products of random matrices*. Duke Math. J.,**45**:311–349.MathSciNetCrossRefGoogle Scholar - 8.A. N. Kolmogorov (1936):
*Über die beste annaherung von functionen einer functionsklasse*. Math. Ann.,**37**:107–111.MATHMathSciNetCrossRefGoogle Scholar - 9.A. N. Kolmogorov (1956):
*Asymptotic characteristics of certain totally bounded metric spaces*. Dokl. Acad. Nauk SSSR,**108**:385–389 (in Russian).MATHMathSciNetGoogle Scholar - 10.P. P. Korovkin (1969):
*An axiomatic approach to certain questions in the constructive approximation theory of functions of one variable*. Uncen. Zap. Kaliningr. Gos. Ped. Inst.,**69**:91–109 (in Russian).MathSciNetGoogle Scholar - 11.B. B. Mandelbrot (1982): Form, Chance, and Dimension. New York: Freeman.Google Scholar
- 12.Bl. Sendov (1962):
*Approximation of functions by algebraic polynomials with respect to a metric of Hausdorff type*. Godishnik Sofia Univ. Mat. Fak.,**55**:1–39 (in Bulgarian).MATHMathSciNetGoogle Scholar - 13.Bl. Sendov (1969):
*Certain questions in the theory of approximation of functions and sets in the Hausdorff metric*, Uspekhi Mat. Nauk,**24**(5):141–178 (English translation (1969): Russian Math. Surveys,**24**(5):143–183).MATHMathSciNetGoogle Scholar - 14.Bl. Sendov (1990): Hausdorff Approximations. Dordrecht: Kluwer.Google Scholar
- 15.Bl. Sendov (1993):
*Integral Hausdorff distance*. C. R. Acad. Bulgare Sci.,**46**(10):21–24.MATHMathSciNetGoogle Scholar - 16.Bl. Sendov, V. A. Popov (1970):
*On certain properties of the Hausdorff metric*. Mathematica (Cluj),**8**:163–172 (in Russian).MathSciNetGoogle Scholar - 17.V. Veselinov (1988): Approximation of Functions by Linear Operators with Respect to Hausdorff and Uniform Metrics. Ph.D. Thesis, Sofia University (in Bulgarian).Google Scholar