• DocumentCode
    3021732
  • Title

    Statistical evaluation of fractal coding schemes

  • Author

    Hurtgen, B.

  • Author_Institution
    Inst. fur Elektrische Nachrichtentechnik, Rheinisch-Westfalische Tech. Hochschule (RWTH) Aachen
  • Volume
    3
  • fYear
    1995
  • fDate
    23-26 Oct 1995
  • Firstpage
    280
  • Abstract
    This paper reports on investigations concerning the convergence of the reconstruction process in fractal coding schemes from a statistical point of view. For a rather general family of “fractal operators” a necessary and sufficient condition for the convergence based on the spectral radius of the fractal operator is provided. Emerging from this condition the probability density function (PDF) of the magnitude of the eigenvalues is formulated which enables to determine a probabilistic measure for the convergence of the reconstruction process. Since the PDF considerably depends on the structure of the operators, various coding schemes can be analyzed with respect to their convergence properties in a statistical sense. The results presented indicate that certain types of operators are less suited for applications in the field of fractal coding compared to others
  • Keywords
    convergence of numerical methods; eigenvalues and eigenfunctions; fractals; image coding; image reconstruction; probability; PDF; convergence properties; eigenvalues; fractal coding; fractal operator; fractal operators; image reconstruction; necessary condition; probabilistic measure; probability density function; spectral radius; statistical evaluation; sufficient condition; Convergence; Density measurement; Discrete cosine transforms; Eigenvalues and eigenfunctions; Fractals; Iterative decoding; Probability density function; Random processes; Signal processing; Statistical analysis;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Image Processing, 1995. Proceedings., International Conference on
  • Conference_Location
    Washington, DC
  • Print_ISBN
    0-8186-7310-9
  • Type

    conf

  • DOI
    10.1109/ICIP.1995.537632
  • Filename
    537632