PDA

View Full Version : Starting Point?



NVentive
04-02-2007, 09:11 AM
I'm trying to learn more about how the 'data side' of video works. We've got some people in our group who have experience writing voice & music reco software, we're now working on a very basic video search experiment. 'No big plans yet, just playing around, and I'm being asked questions that I cannot begin to answer.
Does anyone know any good places to start trying to understand at least a little of how video is actually represented in code?

Thanks -- S

regularfry
04-05-2007, 06:38 AM
What is it you want to know about? The actual algorithms that convert a block of pixels to an encoded representation and then back again? Are you looking for libraries to handle this, or do you want to get your hands dirty with writing codecs yourself? Or are you more interested in higher-level things, like how you handle streaming to without getting blocked on IO, and that sort of thing?

NVentive
04-05-2007, 04:48 PM
For starters, I need to learn more about how the algorithms generate the blocks of code -- I doubt any of us will ever want to write a codec, and just learning how they work will probably take a while. Once we learn how the code is generated, we will (hopefully) be able to back up and look at the code on a wider level, to see if any of the voice/music recognition experience we already have will be of any use. We know that there are video reco routines out there -- we've seen them in security systems -- but we don't know if we ourselves be able to generate anything useful...

Thanks --

regularfry
04-06-2007, 12:42 AM
Depending on your experience with audio compression, the MPEG-1 spec is probably a good place to start. It's here (http://www.chiariglione.org/mpeg/standards/mpeg-1/mpeg-1.htm). MPEG-1 is a primitive codec by modern standards, but it's still widely used and has a lot of features that are still popular with modern codecs. I believe there are open source implementations for you to follow along at home, but I don't know how legal they are in places where software patents are legitimate. It's also been analysed to death, so there's a lot of literature out there on it.

xfulcrumx
04-06-2007, 01:05 AM
For starters, I need to learn more about how the algorithms generate the blocks of code -- I doubt any of us will ever want to write a codec, and just learning how they work will probably take a while. Once we learn how the code is generated, we will (hopefully) be able to back up and look at the code on a wider level, to see if any of the voice/music recognition experience we already have will be of any use. We know that there are video reco routines out there -- we've seen them in security systems -- but we don't know if we ourselves be able to generate anything useful...

Thanks --

here -- http://forums.microsoft.com/MSDN/default.aspx?SiteID=1

or -- http://en.wikipedia.org/wiki/Multimedia_compression

or --
http://en.wikipedia.org/wiki/Data_compression

or--
http://ict.ewi.tudelft.nl/index.php?option=com_sections&id=60&Itemid=124

NVentive
04-06-2007, 04:08 PM
Thanks guys --

I had done some searches, obviously hadn't looked in the right places....

Happy Easter to all -- S


Stephen Self
www.n-ventivetv.com

Bog
04-06-2007, 04:33 PM
I can't help it! I just can't!

0,0,0 t=0

There. Starting point.








Sorry!

NVentive
04-07-2007, 03:05 PM
ugh....

I guess there might be a t for 2 pun out there somewhere, perhaps I should not tempt anyone.....

;-) 2 all -- Happy Easter




Stephen Self
www.n-ventivetv.com