| Author |
Message |
amith
Joined: 17 Feb 2005
Posts: 42
|
 16 Feb 2006 11:09 compression ratio |
|
|
|
Hi ,
what does this implies , compression ratio of 25:1
and is DCT is lossy or lossless , before quantization.
|
|
| Back to top |
|
 |
msemsar
Joined: 17 Feb 2006
Posts: 37
Helped: 4
Location: Iran
|
| 17 Feb 2006 13:49 Re: compression ratio |
|
|
|
Hi,
Compression ratio means the amount of compression in order to coding a raw data, in other words when we say the compression ratio is 25:1 it means that the size of coded stream is 1/25 of the original one, in such case you can convert a 100kB file to a 4KB one,
About the DCT, because DCT coefficients are fractional and not integer, so there is loss when we do inverse DCT, but I thinks the amount of loss is negligible.
Best regards,
Mehdi
|
|
| Back to top |
|
|
sinu_gowde
Joined: 10 Nov 2005
Posts: 147
Helped: 21
|
| 20 Feb 2006 9:19 Re: compression ratio |
|
|
|
Hi Amith,
Compression Ratio:
The compression ratio[/colo] [color=darkred]is a single number that can be used to predict the performance of any internal-combustion engine. It is a ratio between the volume of a combustion chamber when the piston is at the bottom of its stroke, and the volume when the piston is at the top of its stroke. The higher the compression ratio, the more mechanical energy an engine can squeeze from its air-fuel mixture. Higher compression ratios, however, also make detonation more likely.
for more information:
* http://en.wikipedia.org/wiki/Compression_ratio
&
compression ratio of 25:1 - This means that, the algorithm used to compress the data, had compressed it to this extent. it has compressed 25 data elements to a single data element.
cheers...
|
|
| Back to top |
|
|
ArjunSC
Joined: 24 Dec 2005
Posts: 50
Helped: 2
|
| 22 Feb 2006 8:31 Re: compression ratio |
|
|
|
hi amith,
DCT is lossy compression and DFT is not...........quantization is also a lossy process.
compression ratio is the sizeoffile(before compression)/sizeoffile(after compression).
|
|
| Back to top |
|
|
amith
Joined: 17 Feb 2005
Posts: 42
|
| 24 Feb 2006 6:39 Re: compression ratio |
|
|
|
Hi Arjun,
Thanks for the information ,but
while reconstructioning an DCT Image the values is slightly differes which is neglible , but how DFT is not lossy can u clear me .does it give with out error , in reconstruction.
|
|
| Back to top |
|
|
hwangje3
Joined: 08 Jan 2008
Posts: 3
|
| 08 Jan 2008 7:48 compression ratio |
|
|
|
| If not quantizing after transform, it'll prefectly reconstructed.
|
|
| Back to top |
|
|
shakeelsultan
Joined: 10 Aug 2006
Posts: 16
|
| 25 Jan 2008 16:06 Re: compression ratio |
|
|
|
| Well compression ratio gives a measure of how much your algorithm can compress the data, Moreover DCT is lossless before quantization.
|
|
| Back to top |
|
|
hairo
Joined: 20 Nov 2006
Posts: 126
Helped: 16
Location: nowhere
|
| 29 Jun 2008 17:47 compression ratio |
|
|
|
Hi,
I'm a bit confuse here.
Before quantization, is DCT lossless?
I know that once quantize, it will be lossy compression.
Please anyone clear my doubt.
Thanks.
|
|
| Back to top |
|
|
deepu_s_s
Joined: 24 Mar 2007
Posts: 287
Helped: 10
|
| 03 Jul 2008 21:51 compression ratio |
|
|
|
Hi Hairo,
When DCT is performed on the macroblock , the data is not lost.
All the higher frequencies are grouped and all the low frequencies are grouped in a 8x8 matrix form.
Most of the data is represented by the upper left value.
Here there is no loss because of DCT. We are just representing in a newer domain, spatial domain.
After this, the quantization takes places . It removes the higher frequency values as they are very small or zero's.
Removing these , wont be much loss. But it is still a lossy image..
I hope u understood.
If any doubts... get back to me
Thanks and Regards
Deepak
|
|
| Back to top |
|
|
hairo
Joined: 20 Nov 2006
Posts: 126
Helped: 16
Location: nowhere
|
| 04 Jul 2008 0:04 compression ratio |
|
|
|
Hi Deepak,
Thanks a lot for the explanation.
The point is quantization process makes the DCT a lossy transform.
|
|
| Back to top |
|
|
deepu_s_s
Joined: 24 Mar 2007
Posts: 287
Helped: 10
|
| 04 Jul 2008 16:47 compression ratio |
|
|
|
Yes, But that loss can be neglected.
The loss occurs because , in the quantization u scale the DCT co-efficients. Remember DCT values are of floating point.
So When u do inverse quantization on the decoder side, u may not get the exact result as that of encoder side.
Hence it is a lossy compression. But this can be neglected
Added after 33 seconds:
IF u this i helped u ......... Please clicked on helped me
|
|
| Back to top |
|
|
