Bayer to RGB bicubic algorithm

Discussion in 'Digital Photography' started by Christian, Jan 29, 2004.

  1. Christian

    Christian Guest

    Hi, I implemented the algorithms for converting from a CCD sensor
    (Bayer/RAW data) into RGB. I used the nearest neighbour and the
    bilinear algorithms, but I am not satisfied with the results.

    Does any one knows how to do the Bicubic algorithm?

    The data from the sensor is comming as:

    G R G R ...
    B G B G ...
    G R G R ...
    B G B G ...
    .. . . . ...

    Thanks,
    Christian
    Christian, Jan 29, 2004
    #1
    1. Advertising

  2. Christian

    Chris Brown Guest

    In article <>,
    Christian <> wrote:
    >Hi, I implemented the algorithms for converting from a CCD sensor
    >(Bayer/RAW data) into RGB. I used the nearest neighbour and the
    >bilinear algorithms, but I am not satisfied with the results.
    >
    >Does any one knows how to do the Bicubic algorithm?


    Actual production demosaicing algorithms are more sophisticated than any of
    these approaches, and will get significantly better results. Check out the
    source code for Dave Coffin's dcraw for more info:

    http://www.cybercom.net/~dcoffin/dcraw/
    Chris Brown, Jan 29, 2004
    #2
    1. Advertising

  3. Christian

    Guest

    (Christian) wrote:

    > Does any one knows how to do the Bicubic algorithm?


    A description of many algorithms:

    http://www-ise.stanford.edu/~tingchen/

    The best ones are either hideously expensive in CPU, closely guarded
    trade secrets, or under patent protection. The latter case must be
    carefully investigated if you ever decide to release code.
    , Jan 30, 2004
    #3
  4. Christian

    Christian Guest

    I am trying some of the algorithms from Chen, but I can't figure out
    the bi-cubic one. Is there any paper about that one? code?

    Thanks!

    wrote in message news:<>...
    > (Christian) wrote:
    >
    > > Does any one knows how to do the Bicubic algorithm?

    >
    > A description of many algorithms:
    >
    > http://www-ise.stanford.edu/~tingchen/
    >
    > The best ones are either hideously expensive in CPU, closely guarded
    > trade secrets, or under patent protection. The latter case must be
    > carefully investigated if you ever decide to release code.
    Christian, Jan 30, 2004
    #4
  5. Christian

    Guest

    (Christian) wrote:

    > I am trying some of the algorithms from Chen, but I can't figure out
    > the bi-cubic one.


    Use the equation that Chen provides. It is applied to each colour
    plane independently (which happens to be the main reason why it
    doesn't work as well as one might hope). The cubic technique is just
    like the bilinear one, except a cubic is used as the "kernel". For
    code, you might be able to extract something useful from ImageMagick
    (which has this interpolator, as well as a long list of other ones),
    but you'll probably be able to code something up on your own alot
    faster.
    , Jan 31, 2004
    #5
  6. On Thu, 29 Jan 2004 17:26:29 +0000, Chris Brown wrote:

    > In article <>, Christian
    > <> wrote:
    >>Hi, I implemented the algorithms for converting from a CCD sensor
    >>(Bayer/RAW data) into RGB. I used the nearest neighbour and the bilinear
    >>algorithms, but I am not satisfied with the results.
    >>
    >>Does any one knows how to do the Bicubic algorithm?

    >
    > Actual production demosaicing algorithms are more sophisticated than any
    > of these approaches, and will get significantly better results. Check out
    > the source code for Dave Coffin's dcraw for more info:
    >
    > http://www.cybercom.net/~dcoffin/dcraw/


    The code in dcraw implements VNG interpolation, which is both very
    high quality, and patent unencumbered, as far as I'm aware.

    Mike.
    Mike Brodbelt, Feb 2, 2004
    #6
    1. Advertising

Want to reply to this thread or ask your own question?

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Bicubic Interpolation

    , Mar 3, 2006, in forum: Digital Photography
    Replies:
    4
    Views:
    1,336
    Matt Ion
    Mar 4, 2006
  2. Lee
    Replies:
    35
    Views:
    1,021
    John McWilliams
    Jun 12, 2006
  3. John Navas

    Adobe Bicubic Smoother upsize vs Genuine Fractals

    John Navas, Nov 27, 2007, in forum: Digital Photography
    Replies:
    30
    Views:
    2,175
    Bill Tuthill
    Dec 1, 2007
  4. Morton

    Colorsmart RGB or Adobe RGB?

    Morton, Dec 13, 2007, in forum: Digital Photography
    Replies:
    2
    Views:
    3,617
    Morton
    Dec 14, 2007
  5. Alfred Molon

    Lanczos or Bicubic better for upsizing?

    Alfred Molon, Feb 9, 2008, in forum: Digital Photography
    Replies:
    8
    Views:
    19,569
    Allodoxaphobia
    Feb 10, 2008
Loading...

Share This Page