Re: 3d fourier analysis for movie compression?
- From: Thomas Richter <thor@xxxxxxxxxxxxxxxxxxxxxxxxxx>
- Date: 27 Apr 2006 12:03:50 GMT
Hi,
I have an idea regarding compression of movies. Please, tell me if it makes
sense.
Movie can be treated as dynamically changing two dimensional picture or as a
static three dimensional picture (when time is z axis, you know). Hm, so
maybe one could do the three dimensional Fourier analysis of such 3d
picture? And then compress it just like JPEG? Maybe it would work nice?
Of course you can. There are in fact some coding schemes that try something
like it. IIRC, Pegasus had or still is offering a video codec that performs
something like a wavelet transformation in the time axis - which works fine
for surviellience applications where you typcially have no or little motion.
JPEG2000 part 2 can do that as well (actually, one of my test sequences is
taken from a movie and tests wavelets in the time axis).
However, none of these coding schemes outperform the video codes that have
been especially designed for motion picture coding, i.e. H264 and related.
All of them use block-based approaches for motion compensation. The reason
lies in the nature of the images: Moving pictures typically contain a
collection of well-separate objects that move independently.
So long,
Thomas
.
- References:
- 3d fourier analysis for movie compression?
- From: piotr_sobolewski
- 3d fourier analysis for movie compression?
- Prev by Date: Re: Semi-new Compression Method Fully Implemented
- Next by Date: [Discussion] Image Compression using Independent Component Analysis (ICA)
- Previous by thread: Re: 3d fourier analysis for movie compression?
- Next by thread: Implementation of LZS algorithm for user space code
- Index(es):
Relevant Pages
|