522446 v2 basic functions

1.1 Functions:

• Diff()
• Trapz()
• Cumtrapz()

Cumtrapz(1/Fs, sig.^2)

Cumtrapz(Y, dim)

- where Y = time dimension = 1 / frequency

• Dim = amplitudes

Making time vector t = (0:length - 1) / Fs

1.3 Functions:

• Snr()

• (power) --> mean(x.^2)

• SNR = 10*log10(power signal / power noise)

Signal = (index: index)

Noise = (index:index)

Powersignal = (mean(sig.^2))

Powernoise = mean(noise.^2)

1.4 Functions

• Always use 4th order filter

• Check cutoff freq by plotting

• Rectify signal --> abs(sig)

• Filter rectified signal --> butter() ---> filtfilt()

[b,a] = butter(4,4/(Fs/2), '[filter type]')

Env1 - filtfilt(b,a,abs(sig1))

Env2 = filtfilt(b,a,abs(sig2))

1.5 Functions

• Downsample()

• Interp1()

Downsample factor = Fs_original / Fs_downsample

Downsample(sig, downsample factor)

Interp1(x,y,t,'[interp method]')

Where x = downsample time vector (1/Fs)

Where y = signal downsampled

Where t = time vector with time values that don't have y value

2.2 Functions

• Sig = sig-mean(sig)

• Lowpass filt

OR

• DC OFFSET = 0 Hz signal

• Make cutoff (0.1 --> x)

• Bandpass filt

Mean(diff(t)) = Fs

Fc = [0.1 100]/(Fs/2)

DC OFFSET REMOVAL

• Sig = sig-mean(sig)

Noise removal (lowpass)

[b,a] = butter(4, 10/(Fs/2), 'low')

Sigfilt = filtfilt(b,a,sig)

2.3 Functions

• iirnotch()

Wo = 60/(Fs/2)

Bw = 2/(Fs/2)

60 Hz for North America

Wo = noise freq

Bw = bandwidth

2.4 Functions

• EMG, ECG, EEG

• EMG: freqs up to 400-500 Hz

• ECG/EEG: freqs up to 50 Hz

• ECG: Large peak at 1-2 Hz (heartbeat)

• EEG: no large peak at 1-2 Hz

[John D'Errico]

What is your question (concerning MATLAB) in this?