in reply to Re: GIF Image Processing
in thread GIF Image Processing

Please correct me if I'm wrong (it happens all the time) but GIF images are not compressed as such, instead they use an index map into a pallette of colours.

GIF uses the Lempel-Ziv-Welch (LZW) compression algorithm, for which Unisys has a patent. There's a reasonably good explanation of the compression scheme here.

Once uncompressed, you get a matrix of indices into a color palette.

Replies are listed 'Best First'.
Re: Re: Re: GIF Image Processing
by gri6507 (Deacon) on Jan 28, 2002 at 10:15 UTC
    You are correct in your statement - GIFs do use LZW compression. However, ideally I want to use an image to detect its edges in as few clock cycles as possible. I thouhgt that if "decompression" was skipped, then I could achieve this goal (or at least come to a rough approximation of the edges).
      As long as the image is compressed, there is no way to get useful information about any of it without decompressing it. Ideal compression looks like white noise and is about as informative.

      The decompression step is not optional.

      Are you using GIF because it's an outside decision or is it becaue of the compression? If you have a choice, you may be able to find a format that would better fit your needs.

      -Lee

      "To be civilized is to deny one's nature."