Re: [Tftb-help] Urgent Help needed on wavelet (fwd)

[Eric Chassande-Mottin]

> Thanks for your prompt reply. I have some ideas on what you are talking 
> about and I started agree with you. I have re-read the source that I 
> have got, and I realise that the "tau" is defined as the time scale in 
> his defintion, which I think is different than "a" in euqation 1 which 
> equals to "fo/f" or "f/fo", and thus results a different 
> multi-resolution features.

Ok, I understand your point, now. I was wrong yesterday when I said that 
tfrrmsc computes exactly W(tau,omega). tfrrmsc computes a time-scale
representation (t,a) like in the Eq (1) of your previous message.
you want a 1/scale-frequency representation.

> I have tried to compute my work using tfrrmsc, and it results a good 
> time resolution at high frequency region, and poor time resolution when 
> frequency is low.
>
> Another question is, is there anyway for me to compute the 
> multi-resolution properties like what the paper describe? Do I need to 
> go into the m file and change it?

yes, you'll have to. I would extract from tfrrmsc.m the part of the code 
which calculates tfr and adapt the loops so that you implement Eq (3).

> W(tau,omega) = (tau^-0.5) x Integrate [f(t) x h(t/tau) x exp(j*omega*tau)] dt 
> (3)

eric.