From: Summer Dan on
Hi,

I am using a lowpass FIR filter in Matlab in a video of mine to cut some frequencies and get a curve of the videos RGB values for each frame. The sampling rate is 24 frames/sec or 0.04. I set my cutoff frequency to 0.04 and the curve looks pretty smooth, but when I use a lower value cutoff, i.e 0.01 the curve is perfectly smooth. The higher the value the more noise it keeps.

But shouldn't the cutoff frequency be higher than the sampling rate, i.e >0.04? How can I explain this?

Thank you
From: Arthur Thompson on
"Summer Dan" <summerdan(a)ymail.com> wrote in message <i43mac$5hq$1(a)fred.mathworks.com>...
> Hi,
>
> I am using a lowpass FIR filter in Matlab in a video of mine to cut some frequencies and get a curve of the videos RGB values for each frame. The sampling rate is 24 frames/sec or 0.04. I set my cutoff frequency to 0.04 and the curve looks pretty smooth, but when I use a lower value cutoff, i.e 0.01 the curve is perfectly smooth. The higher the value the more noise it keeps.
>
> But shouldn't the cutoff frequency be higher than the sampling rate, i.e >0.04? How can I explain this?
>
> Thank you

I think I can explain this. By setting your cutoff frequency lower than your sample rate/2 you are throwing away the high-frequency components of your data. Of course, this also attenuates the noise more strongly. Apparently you just need to tune your filter to strike the correct balance between blur and noise.

Hope that helps,
Arthur
 | 
Pages: 1
Prev: function outputs
Next: solving 2 functions