Re: Byte difference pre-processing, any advantage?

<collection60@xxxxxxxxxxxxxx> wrote in message
news:1179329800.149631.314280@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Let's say I was going to try experimenting to see if doing byte math
differences on strings, to see if there was any advantage to be
gained.

Anyone know if this will help or not?

I kind of assume that, seeing as byte-difference isn't really used to
help create more patterns, that this technique doesn't really help!
But then I've not seen anyone saying it doesn't.

Basically, let's say you have a string like this:

ABCDEFE

You could store the first byte plain: 65
The next byte would be: 1
And all the rest would be: 1
The last byte would be: -1 though, or perhaps 255, using a numeric
wrap-around algorithm.

The nice thing about this pre-processing is that it's very easy and
quick to perform.

I'm wondering though, would it help increase the number of patterns
available, in the same manner that BWT does?

For what it's worth, a similar lossless concept called BOCU is used to
compress UTF-8 strings. This one actually DOES decrease the size of
the result, though :)

While the technique may or may not be helpful for text, it would certainly
be helpful as a preprocessor for contone images through some compression
algoritms, particularly for synthetic images such as computer-generated
images or prints.


.

Relevant Pages