bwmod — arithmetically modify pixel values in black and white bw file
bwmod [-c ] [ -a add -s sub
-m mult -d div -A
-e exp -r root
-S shift -M and
-O or -X xor
-R -t
] file.bw > file2.bw
bwmod
will perform a series of arithmetic operations on each 8-bit pixel
of its input/output stream. This is a
bw(5)
file format.
The pixel values range from 0 to 255 and are internally operated on
in floating point form (unless the
-c
option has been specified) so that dynamic range, truncation, and sign
are not a problem. On output they are converted back to 8-bit unsigned
values, with clipping to 0 and 255 for values less than 0 and greater than 255. If any pixels are clipped,
the numbers of pixels clipped, low and high, are reported to standard error on completion.
If the
-c
option is specified, then intermediate results are stored as signed characters, though
the command line arguments remain in floating point
format.
This is useful when one wishes to take advantage of wrap-around
characteristics of binary mathematics.
Any number of operations can be given and they are applied in order.
The choices are:
-a val
to add a value.
-s val
to subtract a value.
-m val
to multiply by a value.
-d val
to divide by a value.
-A
to take absolute value.
-e val
to exponentiate, i.e. raise to the power
val.
-r val
to take the
valth
root.
-S val
to shift by val bits. If
val is positive, this uses C-language "^<<" operator, doing left-shift; if
val is negative, this uses C-language ">>" operator, doing right-shift by
-val bits. This is a bitwise operation,
and if it is about to be used, both the pixel value and val
are truncated to integer. Unless -c is in effect,
the resulting new value will again be placed in floating-point. (So it's "user beware" if this option is applied to a
non-integer value.)
-M val
to do "and" operation with val. See -S for "user beware" about bitwise operation.
-O val
to do (inclusive) "or" operation with val. See -S for "user beware" about bitwise operation.
-X val
to do "exclusive or" operation with val. See -S for "user beware" about bitwise operation.
-R to do rounding to the nearest integer.
-t to do truncation to integer; notice the automatic use of truncation for the bitwise options
(-S, -M, -O, and -X).
All arguments can be floating point values. Note that this command can be applied to color pix(5) files also, but no distinction can be made between color planes.
The command
bwmod -m-1 -a255 < file.bw > file2.bw
will produce the negative of a file by first multiplying by minus one, and then adding 255 to shift the pixels back into the 0->255 range. (I.e., replace each pixel value by 255 minus that value.)
One can apply a gamma factor to a file as follows:
bwmod -d255 -r2.2 -m255 < file.bw > file2.bw
which will first normalize the pixel values (i.e., put them into 0->1, noting the internal use of floating point), then take the 2.2 root of each value, and finally scale the pixel values back to 0->255. [Note however that applying gammas this way is not recommended since a function call per pixel results!]
Suppose you ran bwstat on a file and found min=80, max=210, mean=100.
If you wish to expand the range of intensities, while still preserving
the mean, this can be done by:
bwmod -s100 -m1.409 -a100 < file.bw > file2.bw
where 1.409 results from (255.0-100.0)/(210.0-100.0), which is what one would use
to boost the max value (210) to full scale (255) after subtracting the
mean (100). Subtracting and adding 100 makes the multiply occur
"about that point"; i.e., values less than 100 get smaller, and those
above get higher. A file.bw for this example can be created with these three command lines:
gencolor -r 11 80 >file.bw
gencolor -r 262131 100 >>file.bw
gencolor -r 2 210 >>file.bw