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.
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.
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.
In addition to the options common to all programs based on libnetpbm
(most notably -quiet, see
Common Options
), pgmkernel recognizes the following
command line options:
This must be a positive real number.
The default is 6.0.
Default is 255.
This option was new in Netpbm 10.65 (December 2013). Before that, the
maxval is always 255.
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.
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.
Alberto Accomazzi (alberto@cfa.harvard.edu).