Unbiased estimation of the accuracy of digital discretely colored images. (English. Russian original) Zbl 0991.62042

Belyaev, Yu. K.

Theory Probab. Math. Stat. 63, 13-20 (2001); translation from Teor. Jmovirn. Mat. Stat. 63, 13-20 (2000).

The author considers the problem of finding unbiased estimates of congruence accuracy of obtained images to ideal non-observed images. The digital images consist of pixels. Every pixel is defined by a triple \((i,j,c)\), where \(i,j\) define the position of a pixel and \(c\) defines its color. The number of colors is finite. For coloring of pixels of observed images the classifiers with known probabilities of wrong choice of color are used. The obtained images can be made better by re-coloring admitting unbiased estimates of congruence accuracy of re-colored images to the ideal non-observed images.

Reviewer: A.D.Borisenko (Kyïv)

MSC:

62H35	Image analysis in multivariate analysis
62H30	Classification and discrimination; cluster analysis (statistical aspects)
62F10	Point estimation
62G09	Nonparametric statistical resampling methods

Keywords:

unbiased estimation; accuracy; digital discretely colored images; pixels; classifiers; re-coloring