logarithm

Name

logarithm, fftDisplay - Take logarithm of BIFF band (compress dynamic range)

Syntax

 #include <xite/arithmetic.h>

 int logarithm( IBAND in_band, IBAND out_band,
    double eps, double offset, double scale,
    double minimum );

 #include <xite/fft.h>

 int fftDisplay( IBAND in_band, IBAND out_band,
    double scale );

Description

Takes the logarithm of the magnitude of the pixelvalues of a real or complex band according to the algorithm

 out_pix = (log10(eps + |in_pix|) + offset) * scale
 out_pix = (out_pix >= minimum) ? out_pix : minimum

This mapping will compress the dynamic range of an image and make small pixelvalues better visible (e.g. used to display Fourier transform magnitude).

eps equal to DOUBLE_MIN gives the default value

 eps  = 1 / (10 * xsize * ysize)

offset equal to DOUBLE_MAX gives the default value

 offset = -log10(eps)

scale equal to 0.0 gives the default value 20.0.

minimum equal to DOUBLE_MAX gives the default value -10.0 * scale.

For an integral output pixeltype, the resulting pixel-values are rounded. The pixelvalues are limited to fit within the range of the output pixeltype.

fftDisplay is retained for compatibility with previous XITE versions. It is a macro call to logarithm with eps equal to DOUBLE_MIN, offset equal to DOUBLE_MAX and minimum equal to -DOUBLE_MAX. This will give the default values for eps and offset mentioned above. The call to logarithm is followed by a call to shift_band.

Restrictions

Input band may be any pixeltype, output may be double, real or unsigned byte.

Return value

 1 - bands have different xsize
 2 - bands have different ysize
 3 - out_band has wrong pixeltype
 4 - in_band has unknown pixeltype

Author

Trond Sørensen (wrote fftDisplay).

Modified

Svein Bøe (added output pixeltype real and double, fixed bug for input pixeltype real, added lots of more input pixeltypes) when converting from fftDisplay to logarithm.

Id

$Id: logarithm.c,v 1.42 1997/05/27 14:44:03 svein Exp $