View Full Version : Are "ringing" and "mosquito noise" the same? (re. JPEG/MPEG)
shae
12th December 2006, 23:36
I wonder... I always tended to use the term "ringing" to describe what some (or more) call "mosquito noise".
My impression was that the term comes from audio compression (which predates JPEG?) in which it literally means (sonic) ringing, and the same term was later used for the graphical manifestation of the same distortion.
As for the distortion following along contrasted edges, the one that looks like a sharpening artifact, I typically call it haloing.
Is that incorrect?
Chainmax
12th December 2006, 23:40
If I'm not mistaken, both are instances of the Gibbs phenomenon (http://en.wikipedia.org/wiki/Gibbs_phenomenon) (which includes haloing and JPEG-like "grain") but are different kinds of artifacting.
Mug Funky
13th December 2006, 00:12
yeah. ringing is the consequence of any kind of frequency based filtering (esp steep low/highpassing and notch filtering). happens in audio, analog video, image processing (sometimes :)).
mosquito noise is the same effect of removing frequencies but contained within the DCT blocks - hence it appears around edges when there's too much quantization in that block.
shae
13th December 2006, 00:23
So ringing isn't mosquitoing, and it is haloing. Aye?
And are video and audio ringing the same type of artifact?
reepa
13th December 2006, 03:35
Ringing an artifact usually associated with resampling algorithms. Since resampling can be done on both video and audio, the artifacts are basically the same. Here's an image with ringing artifacts (probably caused by a bad resizing algorithm):
http://modeemi.cs.tut.fi/~leopold/AV/ResolutionComparison/178_NTSC_DVD_133_2.jpeg
I've never seen ringing caused by JPEG compression. Mosquito noise looks different, and is caused by JPEG compression. So they aren't the same thing.
foxyshadis
13th December 2006, 05:58
It's hard to come up with a good name for the artifact - "high frequency dct quantization artifacting" is quite a mouthful, and mosquito noise doesn't sound very scientific - so I guess ringing just stuck, given they occur in the same areas and have a similar visual effect at first glance.
Subsampled wavelet codecs show real ringing, but that comes more from the subsampling than the wavelet noise.
chipzoller
4th January 2007, 21:25
Then what would be some of the better (or at least more popular) avisynth plugins to combat either or both of these phenomena?
And also, would ringing/mosquito noise accompany blocking inherent to satellite (low MPEG2 bitrate) captures?
Are deblocking filters also de-mosquito/ringing filters as well?
I'm trying to learn to clean up both of these artifacts from my satellite captures (http://www.megaupload.com/?d=SFKHF9ZF) and so far I can't find a filter combination to eliminate both of them in conjunction-only one or the other, and when removing one it seems to intensify the other. Any suggestions for eliminating these types of noise?
Soulhunter
6th January 2007, 04:48
I've never seen ringing caused by JPEG compression...
Hmm...
http://img170.imageshack.us/img170/5353/samplerz1.png (http://imageshack.us)
Btw, we already had this discussion more than 1 year ago! (http://forum.doom9.org/showthread.php?t=99795) ;D
Bye
Blue_MiSfit
6th January 2007, 10:35
Satellite (especially Dish network like I have at home) is absolutely dreadful in terms of compression quality. It's been through so many 1 pass hardware MPEG-2 encoders, and is given so little bitrate ultimately that the quality is just terrible.
Honestly, it's better than analog cable, but not by much, and is (to my eyes) unwatchable on a revealing device like a large LCD HDTV. Some channels also have EXCESSIVELY strong temporal filtering to the point of reducing any motion to a smeared blur. Most of the time it's just lots of blocks, generally washed out details (and colors actually), and some ringing. Really bad. The HD stuff is better because the smaller macro blocks help cover things up, but it's still a shame the transponders don't have more bandwidth!
chipzoller
6th January 2007, 15:46
That's one thing that I'm finding more true every day. You really know it's bad when you're 20 feet away from a TV and still can see macroblocking! Also the aliasing is terrible! And I don't know why it's that bad but anytime there are any sort of linear patterns on the screen it just looks like a saw blade.
reepa
7th January 2007, 18:25
Hmm...
http://img170.imageshack.us/img170/5353/samplerz1.png (http://imageshack.us)
Btw, we already had this discussion more than 1 year ago! (http://forum.doom9.org/showthread.php?t=99795) ;D
Bye
I'll quote myself from the earlier post: "Mosquito noise looks different, and is caused by JPEG compression."
IMO that is an example of mosquito noise, not ringing. For an example of ringing, see my earlier post.
Soulhunter
7th January 2007, 18:59
Isnt ringing simply a "echo" of the high frequencys, so edges and fine details? At least thats how I understand it... But if I had to differentiate between mosquito noise and ringing then the dark dots [which look like mosquitoes, especially in motion...] are what id describe as mosquito noise... The vertical edge echo on the other hand id just define as ringing... And btw, your sample pic shows the same, echoes of the edges, how they are produced doesnt matter imo!
Bye
dadix
31st January 2012, 10:43
Mosquito noise : http://www.pcmag.com/encyclopedia_term/0,2542,t=mosquito+noise&i=55914,00.asp
Blue_MiSfit
1st February 2012, 01:07
@dadix: Please don't resurrect 5 year old threads.
dadix
1st February 2012, 10:11
@Blue_MiSfitIf if you give me a script to delete this type of artifact(in 16 bpp) i will, if not ... well ... i will spin around here:)
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.