Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.compression > #3117

Re: Lempel Ziv compression implementation

Show all headers | View raw

mathieu <mathieu.malaterre@gmail.com> wrote:
> On Saturday, July 25, 2015 at 10:28:20 PM UTC+2, robert...@yahoo.com wrote:
(snip) 
>> Page 11 seems to be clear that it's just LZW as described in the
>> original Welch paper, per reference #21.
 
> Any implementation out there that support 16bits input. 
> Most implementations I found assume 8bits input.

LZW finds repetitions in the data.  Repetitions in 16 bit data are
also repeitions in 8 bit data, so it will still find them.

LZW initializes the dictionary with all 8 bit values, which isn't
so bad.  If you made one that initialized with all 16 bit values,
it would have to start with 17 bit codes, which could be a lot
less efficient.

LZW also works well for smaller data, for example files full of
binary (ASCII '0' and '1') data, even if the actual 0's and 1's
are random, will compress well. It will find strings of 1's and
0's and code for them.

-- glen

Back to comp.compression | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

Lempel Ziv compression implementation mathieu <mathieu.malaterre@gmail.com> - 2015-07-25 06:30 -0700
  Re: Lempel Ziv compression implementation Robert Wessel <robertwessel2@yahoo.com> - 2015-07-25 15:28 -0500
    Re: Lempel Ziv compression implementation mathieu <mathieu.malaterre@gmail.com> - 2015-07-28 03:55 -0700
      Re: Lempel Ziv compression implementation glen herrmannsfeldt <gah@ugcs.caltech.edu> - 2015-07-28 17:41 +0000
      Re: Lempel Ziv compression implementation Dave Brown <0ntariokingston@gmail.com> - 2015-08-04 14:11 -0700

csiph-web