Any data compression experts in here?

[heavyiron8]

I'm a computer science major, and currently taking a course that's basically going to teach lots of data compression and cryptography.

Anyone here an expert in these fields? I'm pretty interested. I might consider this my area of research when I hit grad school.

 

[Cogman]

I am by no means an expert, but I've looked into it. I'll be happy to share what I know.

 

[Ken g6]

Ditto.

One thing I know about the data compression field is that it's heavily patent-encumbered. That is, there are a lot of software patents on compression techniques, and if you develop a new one you're likely to run foul of one of them. Unless you're lucky enough to be in Europe or something.

 

[Leros]

I know there is a mathematical bound on compression. How close are we to this?

 

[Cogman]

I know there is a mathematical bound on compression. How close are we to this?

There isn't an absolute mathematical bound. There is, however, an equation related to the randomness of a given input that dictates the amount of resources it needs http://en.wikipedia.org/wiki/Kolmogorov_complexity

The problem is that randomness is hard to truly define. For example, Pi may look fairly random, however a fairly small program can generate enough digits of it to fill whatever HD you have. That could be considered some pretty awesome compression.

Another example.
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
is really easy to compress, you could say
20a and viola, you have a highly compressed string. (ok, it may not be 20 a's, I didn't want to count it )

 

[TecHNooB]

is this compression in CS land where you only consider bits or are you interested in compressing things using signal processing techniques as well?

 

[heavyiron8]

is this compression in CS land where you only consider bits or are you interested in compressing things using signal processing techniques as well?

I would say both. I'm more interested in the software aspect of it though.

Signal Processing is more associated with double E people I'd say.

 

[Cogman]

I would say both. I'm more interested in the software aspect of it though.

Signal Processing is more associated with double E people I'd say.

signal processing is however image compression borrows a lot from it. Pretty much all lossy compression is based off of the dct which is pretty closely related to signal processing.

 

[Modelworks]

my favorite form of compression to use on embedded hardware is look up tables and substitution . Code runs fast even on slow processors. If you are good at math then compression and cryptography is an awesome field. I know someone in the field but he majored in mathematics not computing though he has picked up some computing skills after he got the math degrees.