Provided by: netpbm_11.01.00-2build1_amd64
NAME
pgmkernel - generate a convolution kernel
SYNOPSIS
pgmkernel [-weight=number] [-maxval=] {size | width height} Minimum unique abbreviation of option is acceptable. You may use double hyphens instead of single hyphen to denote options. You may use white space in place of the equals sign to separate an option name from its value.
DESCRIPTION
This program is part of Netpbm(1). pgmkernel generates a convolution kernel that you can use with pnmconvol. The kernel is one where the weight of each location is inversely proportional to its distance from the center of the kernel. pgmkernel generates a PGM image of width and height size if you specify one argument, or width width and height height if you specify two arguments. pgmkernel computes the convolution function K as follows. K(i,j) = 1 / ( 1 + w * sqrt(i^2 + j^2)) where w is a coefficient specified via the -weight option. i and j are measured in pixels. K is zero everywhere beyond the specified kernel width and height. The sample values in the PGM output have this value scaled and biased using the protocol pnmconvol specifies for representing the real numbers K in PGM.
OPTIONS
In addition to the options common to all programs based on libnetpbm (most notably -quiet, see Common Options ⟨index.html#commonoptions⟩ ), pgmkernel recognizes the following command line options: -weight=number The distance from the center is weighted by this; a higher number means the value falls off more quickly as you go away from the center. This must be a positive real number. The default is 6.0. -maxval=maxval The maxval for the PGM kernel. Default is 255. This option was new in Netpbm 10.65 (December 2013). Before that, the maxval is always 255.
LIMITATIONS
The computation time is proportional to width*height. This increases rapidly with the increase of the kernel size. A better approach could be using a FFT in these cases.
HISTORY
Before Netpbm 10.65 (December 2013), the output was always in Plain (text) PGM format. (Now, like standard Netpbm programs, the default is raw PGM and you can get Plain PGM with a -plain option). Before Netpbm 10.65 (December 2013), this manual said negative values for -weight were valid (as long as they were greater than -1.0). But the program never worked with negative numbers and it isn't clear that the result would be useful, so -weight is now required to be nonnegative and the program fails gracefully if you specify a negative value.
SEE ALSO
pnmconvol(1), pnmsmooth(1) pamgauss(1) pgm(1)
AUTHOR
Alberto Accomazzi (alberto@cfa.harvard.edu).
DOCUMENT SOURCE
This manual page was generated by the Netpbm tool 'makeman' from HTML source. The master documentation is at http://netpbm.sourceforge.net/doc/pgmkernel.html