what is the 'filtfilt' working domain

Is the 'filtfilt' filter works in frequency or time domain?

Answer 1

[Walter Roberson]

It works in the time domain.

[Haya AKKAD]

Thank you very much for your respond.

Please can u briefly clarify for me why the filtfilt working in time domain?

I thought that is work in frequency domain.

And also I have another question, is the "spatial domain" same to "time domain"?

I mean can I say filtfilt works in spatial domain?

[Walter Roberson]

"Zero-phase filtering helps preserve features in a filtered time waveform exactly where they occur in the unfiltered signal."

... from the documentation.

That is, the input is time domain signals and the output is time domain signals.

Or are you asking whether it does an fft and modifies the fft and then does an ifft?

[Walter Roberson]

The filter you pass in can be designed as iir or fir.

Implementation of iir by fft is inefficient. However at least historically filtfilt used fft for fir filters; I do not know if that is still true.

http://matlab.izmiran.ru/help/toolbox/signal/basics20.html

[Haya AKKAD]

Thank u very much.

[Paul]

Is implementation of IIR by fft even possible? In any case, in R2022a at least, filtfilt is implemented with multiple calls to filter

[Walter Roberson]

@Paul

That 2005 Mathworks link describes the algorithm one would need.

[Paul]

IMO, that page is misleading at best and outright wrong at worst. n is too small, even for a FIR filter, for which time domain filtering can be done exactly (in theory) using fft/ifft. However, if n is selected large enough, then I think it can yield a good approximation for stable, IIR filters.