Mercurial > octave
changeset 20658:eafc22e4c857
ntsc2rgb: truncate/scale for all types and performance fixes (patch #8709)
* scripts/image/ntsc2rgb.m: Matlab only accepts input of class double so on
cset 175aed5acb85 it only performed scaling for class double and doing nothing
otherwise. However, this can cause issues for the user since it then it
will need to be constantly aware of the class (the whole point of accepting
any class is not having to worry about it).
In addition, make use of automatic broadcasting (instead of multiplying by
ones to expand), and drop use of find() since it's faster to use the bool
array for indexing than the extra call to find().
| author | Carnë Draug <carandraug@octave.org> |
|---|---|
| date | Sun, 25 Oct 2015 20:44:13 +0000 |
| parents | bc51aa0d604a |
| children | 63f975ff1f7c |
| files | scripts/image/ntsc2rgb.m |
| diffstat | 1 files changed, 3 insertions(+), 7 deletions(-) [+] |
line wrap: on
line diff
--- a/scripts/image/ntsc2rgb.m Sun Oct 25 19:34:56 2015 +0000 +++ b/scripts/image/ntsc2rgb.m Sun Oct 25 20:44:13 2015 +0000 @@ -59,13 +59,9 @@ rgb = yiq * trans; ## truncating / scaling of double rgb values for Matlab compatibility - ## (Other input types are not supported by Matlab. We deal with - ## them anyways, but do not truncate or scale those.) - if (cls == "double") - rgb = max (0,rgb); - idx = find (any (rgb'>1)); - rgb(idx,:) = rgb(idx,:) ./ (max (rgb(idx,:), [], 2) * ones (1,3)); - endif + rgb = max (0, rgb); + idx = any (rgb > 1, 2); + rgb(idx,:) = rgb(idx,:) ./ max (rgb(idx,:), [], 2); rgb = colorspace_conversion_revert (rgb, cls, sz, is_im, is_nd, is_int); endfunction