MAN page from RedHat Other XITE-3.3-3.i386.rpm

LOWPASS,

Section: BUTTERWORTH, (HIGHPASS,)
Updated: BANDPASS,
Index

NAME

lowpass, highpass, bandpass, bandstop, butterworth,exponential, lowpassIdeal, fftMakeLp, fftMakeHp, fftMakeBp,fftMakeBs - make an "ideal", truncated, lowpass filter inthe spatial or Fourier domain

SYNTAX

lowpass [<option>...] <outimage>
highpass [<option>...] <outimage>
bandpass [<option>...] <outimage>
bandstop [<option>...] <outimage>
butterworth [<option>...] <outimage>
exponential [<option>...] <outimage>
lowpassIdeal [<option>...] <outimage>

DESCRIPTION

lowpassmakes a lowpass-filter in the spatial or Fourierdomain. The filter used is an ideal lowpass filtertruncated with a circularly symmetric window functionin the image domain.

highpass,bandpassand bandstopbehave correspondingly.

The circular symmetry is achieved using Huang's method.

butterworthmakes a Butterworth lowpass filter in thespatial or Fourier domain.

exponentialmakes an exponential lowpass filter in thespatial or Fourier domain.

lowpassIdealmakes an "ideal" lowpass filter in the spatialor Fourier domain. This filter is the box-shaped idealfrequency response which corresponds to a non-realizablespatial-domain filter. The spatial domain result (usinginverse fft on the frequency-domain result), is an aliasedversion of the non-realizable spatial-domain filter.

fftMakeLpwas a program similar (but not identical) tolowpassin previous XITE versions. It is retained forcompatibility, but now as a script which invokes lowpass.The old fftMakeLpgave results slightly different from theresults of lowpass.This is due to differences in theimplementation of the window function. The main difference isthat filter_sizewill be reduced by two in lowpass,compared to the old fftMakeLp.So, the new script version offftMakeLpincreases filter_sizeby two before invokinglowpasswith Hamming window. Likewise for fftMakeHp,fftMakeBpand fftMakeBs.

OPTIONS

Not all options are accepted by all the programs.

-c cut_frequency

Only for lowpass,highpass,butterworth,exponentialand lowpassIdeal.

Specified relative to the Nyquist frequency, i.e. acut_frequencyof 1.0 equals half the sampling frequency.Default: 0.2.

-f filter_size, -f filter_order

For lowpass,highpass,bandpassand bandstopthisoption gives the diameter of the filter. Default: 41.For butterworthand exponential,it is the filter order.Default: 5. This option can not be used with lowpassIdeal.

-h high_cut_frequency

Only for bandpassand bandstop.

Specified relative to the Nyquist frequency, i.e. ahigh_cut_frequencyof 1.0 equals half the sampling frequency.Default: 0.7.

-l low_cut_frequency

Only for bandpassand bandstop.

Specified relative to the Nyquist frequency, i.e. alow_cut_frequencyof 1.0 equals half the sampling frequency.Default: 0.2.

-pt pixtyp

Only when result is in the spatial domain.

Pixeltype of filter.

r or 8Real

c or 9Complex

d or 10Double

dc or 11Double complex

Default: 8 (real) for spatial-domain result, 9 (complex)for Fourier-domain result.

-r

Only for highpassand bandpasswhen result is in theFourier domain.

Restore image mean value (without scaling the rest of thefilter correspondingly).

-s

Return filter in spatial domain, not in Fourier domain.Default: Fourier domain.

-t image_title

Image title. Default: Lowpass filter.

-w win_type

This option can only be used with lowpass,highpass,bandpassand bandstop.

r or 0Circularly symmetric rectangular (box, boxcar) windowfunction

b or 1Circularly symmetric Bartlett (triangular) window

hm or 2Circularly symmetric Hamming window

hn or 3Circularly symmetric Hanning (von Hann) window

t or 4Circularly symmetric triangular window function

Default: 2.

-x xsize

Horizontal size of the filter image. Default 512.

-y ysize

Vertical size of the filter image. Default 512.

RESTRICTIONS

For a result in the Fourier domain, xsizeand ysizemustbe a power of two and pixtypmust be complex.

RETURN VALUE

0 - OK
1 - Bad pixel type
2 - Unknown window type
3 - Filtersize larger than bandsize
4 - Error in fft2d
5 - Malloc failed
6 - Bad number of arguments

REFERENCE

Jae S. Lim

"Two-dimensional Signal And Image Processing"Prentice-Hall, 1990

AUTHOR

Trond Muri Sřrensen and Svein Bře

This document was created byman2html,using the manual pages.