Home > Archive > Compression > November 2007 > Max. lossless image compression
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Max. lossless image compression
|
|
| Convenor 2007-11-15, 3:56 am |
| Friends,
Could you inform me the technique which compresses the data to the
maximum level. Also, infrom me the maximum lossless compression
possible with images.
Saravanan
| |
| Phil Carmody 2007-11-15, 7:56 am |
| Convenor <trysaran@gmail.com> writes:
> Friends,
>
> Could you inform me the technique which compresses the data to the
> maximum level.
The best technique is to know exactly what the data is already.
That way you can compress it down to zero bits.
Phil
--
Dear aunt, let's set so double the killer delete select all.
-- Microsoft voice recognition live demonstration
| |
| Thomas Richter 2007-11-15, 7:04 pm |
| Convenor wrote:
> Friends,
>
> Could you inform me the technique which compresses the data to the
> maximum level. Also, infrom me the maximum lossless compression
> possible with images.
This is so broad its really hard to answer. As far as standardized
techniques exist, JPEG-LS performs pretty well. Dependent on image
content, it compresses around 1:2, which is also what you should expect
from proprietary software. One can buy here or there improvements with
more complex algorithms (JPEG-LS is not the best possible, but a pretty
high performing, but low-complexity solution), but not too much.
Technique here is simply a clever prediction scheme (context are pixels
around the current one) that adjusts itself to the image statistics, and
an entropy coder backend.
So long,
Thomas
| |
| Convenor 2007-11-16, 4:00 am |
| On Nov 15, 8:04 pm, Thomas Richter <t...@math.tu-berlin.de> wrote:
> Convenor wrote:
>
>
> This is so broad its really hard to answer. As far as standardized
> techniques exist, JPEG-LS performs pretty well. Dependent on image
> content, it compresses around 1:2, which is also what you should expect
> from proprietary software. One can buy here or there improvements with
> more complex algorithms (JPEG-LS is not the best possible, but a pretty
> high performing, but low-complexity solution), but not too much.
>
> Technique here is simply a clever prediction scheme (context are pixels
> around the current one) that adjusts itself to the image statistics, and
> an entropy coder backend.
>
> So long,
> Thomas
Thank you for the information. In order to estimate a compression
technique, which error metric is best, PSNR, RMSE, SSIM, etc.
| |
| Industrial One 2007-11-16, 9:56 pm |
| On Nov 15, 9:42 pm, Convenor <trysa...@gmail.com> wrote:
> On Nov 15, 8:04 pm, Thomas Richter <t...@math.tu-berlin.de> wrote:
>
>
>
>
>
>
>
>
> Thank you for the information. In order to estimate a compression
> technique, which error metric is best, PSNR, RMSE, SSIM, etc.
PNG is the best.
| |
| Thomas Richter 2007-11-16, 9:56 pm |
| Industrial One wrote:
>
> PNG is the best.
PNG is not a metric, it is a format. And no, JPEG-LS compresses better
(so does JPEG2000 lossless)
So long,
Thomas
| |
| Thomas Richter 2007-11-16, 9:56 pm |
| Convenor wrote:
> Thank you for the information. In order to estimate a compression
> technique, which error metric is best, PSNR, RMSE, SSIM, etc.
PSNR and RMSE are identical, just plotted differently. What "best" is
depends on what you want to target at. For visual performance, SSIM is
clearly closer to that than PSNR, but not the ultimate answer. Other
metrics exist.
So long,
Thomas
|
|
|
|
|