IIR to find minimum, maximum enveloppe of a signal
in [DSP]
|
Prev: Tomorrow they are ready to start the LHC collider. The Earth can be exploded in a 1000 seconds!
Next: Tomorrow they are ready to start the LHC collider. The Earthcan be exploded in a 1000 seconds!
From: dbd on 25 Feb 2010 10:54 On Feb 25, 2:22 am, Jerry Avins <j...(a)ieee.org> wrote: > ... > Before going off on a wild goose chase, define /precisely/ what you mean > by "envelope". You can't compute what you can't define. > > Jerry I'd bet you've seen people compute stuff they couldn't even identify, let alone define. Don't we get questions from them here? Dale B. Dalrymple
From: Sebastian Doht on 25 Feb 2010 14:03 vectorizor schrieb: > Thanks to all for all interesting answers. To refocus the > conversation, the signals being processed are images, so I dont need > to know the future :) , and a running min/max is not sufficient, as > I'm more looking for a smooth enveloppe, rather than one looking like > a step function. > >> Hi, >> You could use a single pole IIR filter with time varying alpha (or 2 >> sets of alpha) to track the maxima. >> y[n]=alpha*x[n] +(1-alpha)*y[n-1]. >> Essentially, if the input x[n] is greater than the previous output >> y[n-1], use a faster alpha and if its smaller, use a smaller alpha. >> This is like a fast attack and slow decay kind of a filter. This can >> be used to track envelope efficiently. > > This is the most interesting answer IMHO. I can see how it could > potentially give a smooth enveloppe. But how to control the alpha? > Could you give me more clues/point me to some reference materials? > > Thanks again > > A If you have a good model for the signals you are expecting the alpha can be updated each step by the help of a kalman filter. If you working with images an edge detector (Canny, Hough etc) is really what you need?
From: Jerry Avins on 25 Feb 2010 15:36 dbd wrote: > On Feb 25, 2:22 am, Jerry Avins <j...(a)ieee.org> wrote: > >> ... >> Before going off on a wild goose chase, define /precisely/ what you mean >> by "envelope". You can't compute what you can't define. >> >> Jerry > > I'd bet you've seen people compute stuff they couldn't even identify, > let alone define. Don't we get questions from them here? Well yes, that's a telling point. Did those calculations do anybody any good? Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
From: vectorizor on 26 Feb 2010 06:56 > Before going off on a wild goose chase, define /precisely/ what you mean > by "envelope". You can't compute what you can't define. Here is an image that shows what I want to achieve: http://img8.imageshack.us/img8/8839/11635847.png the signal is in blue, and the maximum and and minimum enveloppes are respectively in red and black. FYI, I generated the enveloppes with a method that is terribly inefficient, but it achieves good results in terms of quality. Note the maximum enveloppe may be well over the signal, this is because these are 1D projections of a 2D signal. Local extrema may be close to the samples present in the plots, without being seen on the plots. Thanks for the help.
From: vectorizor on 26 Feb 2010 07:03
Hi, > If you have a good model for the signals you are expecting the alpha can > be updated each step by the help of a kalman filter. No, the signal is a medical image, i.e. the signal can be anything at any location. > If you working with images an edge detector (Canny, Hough etc) is really > what you need? No at all, I'm looking for the minimum and maximum enveloppe of the 2D signal. See the plots I've uploaded. A |