The bootstrap mean filter for image restoration

Author

Lam, Chi-Kin

Author_Institution

Naval Air Warfare Center, China Lake, CA, USA

fYear

1993

fDate

1-3 Nov 1993

Firstpage

578

Abstract

A bootstrap mean is calculated as follows. An artificial data set is generated by randomly sampling the original data set. A trimmed mean is then calculated for each of the artificial data set. These steps are repeated many times to produce a set of trimmed means. The bootstrap mean is the average of these trimmed means. The bootstrap mean is a more robust estimate of the true mean and the estimation of error is the usual standard deviation. Recent advances in fast computers make it feasible to calculate the bootstrap mean. A bootstrap mean filter was developed and tested using synthetic data with random noise added. Comparisons to mean, median, and trimmed-mean filters show that the bootstrap mean filter is superior in the removal of random noise and the retention of edge information. Implementation in special purpose hardware of this filter is desirable because of its heavy computational requirement. Some candidate solutions are suggested

Keywords

filtering theory; image restoration; median filters; random noise; artificial data set; bootstrap mean; bootstrap mean filter; edge information; error estimation; image restoration; mean filters; median filters; random noise; standard deviation; synthetic data; trimmed mean; trimmed-mean filters; Additive noise; Computer errors; Degradation; Estimation error; Hardware; Image restoration; Image sampling; Information filtering; Information filters; Low pass filters; Noise reduction; Noise robustness; Pixel; Smoothing methods; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Signals, Systems and Computers, 1993. 1993 Conference Record of The Twenty-Seventh Asilomar Conference on

Conference_Location

Pacific Grove, CA

ISSN

1058-6393

Print_ISBN

0-8186-4120-7

Type

conf

DOI

10.1109/ACSSC.1993.342582

Filename

342582