Accession Number : ADA131394

Title :   Nonparametric Estimation by the Method of Sieves.

Descriptive Note : Final technical rept. 1 Jun 80-31 May 83,

Corporate Author : BROWN UNIV PROVIDENCE RI DIV OF APPLIED MATHEMATICS

Personal Author(s) : Geman,Stuart ; McClure,Donald E

PDF Url : ADA131394

Report Date : Jul 1983

Pagination or Media Count : 62

Abstract : The research project has built a theoretical foundation for using the method of sieves to adapt classical estimation principles such as maximum likelihood and least squares to problems with infinite dimensional parameter spaces. The first results about consistency of cross validated estimators of density functions have been obtained. The method of sieves and the principle of maximum likelihood have been used to develop algorithms for digital image processing. Specific applications include image segmentation, reconstruction methods for tomography, image registration methods for moving objects, and surface restoration algorithms. (Author)

Descriptors :   *Nonparametric statistics, *Estimates, Segmented, Computer applications, Maximum likelihood estimation, Least squares method, Algorithms, Digital systems, Image processing, Tomography, Image registration

Subject Categories : Statistics and Probability

Distribution Statement : APPROVED FOR PUBLIC RELEASE