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

Groups > comp.compression > #1293

Re: Words as well as characters as encoding units

Show all headers | View raw

Mok-Kong Shen <mok-kong.shen@t-online.de> wrote:
> Am 05.05.2012 22:17, schrieb glen herrmannsfeldt:
> [snip]
>> If the distribution changes, though, after not so long LZW throws
>> away the table and starts a new one. Huffman doesn't do that.

> In a couple of textbooks I have, there is no mention of thowing
> away of LZ table with change of distribution. Could you please
> recommend some better literatures? 

I don't know of any specific reference. The usual LZW uses 16 bit
codes, for a 65536 entry table. When the table is full, it throws
away all the table entries and starts a new one. 

One commonly considered case is the .tgz file, compressing
the output of tar. Since it is pretty much a concatenation of 
files, different files could have very different distributions.
I believe that there is a code which will dump the table early,
but normally it runs until it fills, then starts over.

> I don't properly understand your last sentence in this context. 
> (I mean there is the adaptive Huffman.)

Yes, I meant non-adaptive. Huffman is, for example, used for
group 3 fax, with a fixed table based on statistical analyis
of documents. That means no table to send, and on the average
isn't too bad.

-- glen

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

Thread

Words as well as characters as encoding units Mok-Kong Shen <mok-kong.shen@t-online.de> - 2012-05-05 10:26 +0200
  Re: Words as well as characters as encoding units James Dow Allen <jdallen2000@yahoo.com> - 2012-05-05 02:41 -0700
    Re: Words as well as characters as encoding units Mok-Kong Shen <mok-kong.shen@t-online.de> - 2012-05-05 15:06 +0200
      Re: Words as well as characters as encoding units Thomas Richter <thor@math.tu-berlin.de> - 2012-05-05 17:05 +0200
        Re: Words as well as characters as encoding units Mok-Kong Shen <mok-kong.shen@t-online.de> - 2012-05-05 19:42 +0200
          Re: Words as well as characters as encoding units glen herrmannsfeldt <gah@ugcs.caltech.edu> - 2012-05-05 18:33 +0000
            Re: Words as well as characters as encoding units Mok-Kong Shen <mok-kong.shen@t-online.de> - 2012-05-05 21:22 +0200
              Re: Words as well as characters as encoding units Thomas Richter <thor@math.tu-berlin.de> - 2012-05-05 21:31 +0200
                Re: Words as well as characters as encoding units Mok-Kong Shen <mok-kong.shen@t-online.de> - 2012-05-05 22:19 +0200
              Re: Words as well as characters as encoding units glen herrmannsfeldt <gah@ugcs.caltech.edu> - 2012-05-05 20:17 +0000
                Re: Words as well as characters as encoding units Mok-Kong Shen <mok-kong.shen@t-online.de> - 2012-05-05 22:36 +0200
                Re: Words as well as characters as encoding units glen herrmannsfeldt <gah@ugcs.caltech.edu> - 2012-05-05 21:29 +0000
                Re: Words as well as characters as encoding units Mok-Kong Shen <mok-kong.shen@t-online.de> - 2012-05-05 23:52 +0200
                Re: Words as well as characters as encoding units Niels Fröhling <spamtrap@adsignum.com> - 2012-05-08 00:42 -0600
                Re: Words as well as characters as encoding units glen herrmannsfeldt <gah@ugcs.caltech.edu> - 2012-05-08 08:19 +0000
                Re: Words as well as characters as encoding units Niels Fröhling <spamtrap@adsignum.com> - 2012-05-09 15:14 -0600
                Re: Words as well as characters as encoding units James Dow Allen <jdallen2000@yahoo.com> - 2012-05-09 22:17 -0700

csiph-web