DATAMANIP commands


PROGRAM NAME

kwiener - Wiener Filtering (Restoration) in Fourier Frequency Domain

DESCRIPTION

kwiener performs restoration of data in the spatial domain by Wiener filtering in the frequency domain. The idea is to deconvolve the blurring function from the blurred data. A few problems exist in performing this operation, as described in great detail in any good image processing book.

A full implementation of a Wiener filter requires detailed knowledge of the signal and noise power spectra. Many times this information is simply unavailable or is very poorly known. In such situations, it is convenient to assign some constant value to the SNR and use that in the Wiener filter implentation. This is a somewhat extreme restriction of the Wiener filter, but it makes the filter very easy to use and can produce better results than the pseudoinverse filter in kinverse. (Better in the sense of much less ringing).

Note that the input data objects will be treated simply as a field of complex data, presumably obtained by application of an FFT, although the data could be generated by many other means. The output data will be either KCOMPLEX if the input data is any type other than KDCOMPLEX. Otherwise, the output will be of type KDCOMPLEX. Data objects with maps attached are not supported. Mask data is ignored, if present.

REQUIRED ARGUMENTS

-i1
type: infile
desc: source data object (FFT)
-i2
type: infile
desc: PSF data object (FFT)
-o
type: outfile
desc: filtered data object (FFT)

OPTIONAL ARGUMENTS

-w
type: double
desc: Wiener parameter
default: 1
bounds: value >= 0.0

EXAMPLES

SEE ALSO

kinverse(1), kfft(1), kcmplx2real(1)

RESTRICTIONS

REFERENCES


K.R. Castleman, "Digital Image Processing", Prentice-Hall, 1979, ISBN 0-13-212365-7, p281 Eq. 2.
Rosenfeld & Kak, p287.

COPYRIGHT

Copyright (C) 1993 - 1997, Khoral Research, Inc. ("KRI") All rights reserved.