Show Sidebar Hide Sidebar

# FFT Filters in Python

Learn how filter out the frequencies of a signal by using low-pass, high-pass and band-pass FFT filtering.

#### New to Plotly?¶

You can set up Plotly to work in online or offline mode, or in jupyter notebooks.
We also have a quick-reference cheatsheet (new!) to help you get started!

#### Imports¶

The tutorial below imports NumPy, Pandas, SciPy and Plotly.

In [1]:
import plotly.plotly as py
import plotly.graph_objs as go
from plotly.tools import FigureFactory as FF

import numpy as np
import pandas as pd
import scipy

from scipy import signal


#### Import Data¶

An FFT Filter is a process that involves mapping a time signal from time-space to frequency-space in which frequency becomes an axis. By mapping to this space, we can get a better picture for how much of which frequency is in the original time signal and we can ultimately cut some of these frequencies out to remap back into time-space. Such filter types include low-pass, where lower frequencies are allowed to pass and higher ones get cut off -, high-pass, where higher frequencies pass, and band-pass, which selects only a narrow range or "band" of frequencies to pass through.

Let us import some stock data to apply FFT Filtering:

In [2]:
data = pd.read_csv('https://raw.githubusercontent.com/plotly/datasets/master/wind_speed_laurel_nebraska.csv')
df = data[0:10]

table = FF.create_table(df)
py.iplot(table, filename='wind-data-sample')

Out[2]:

#### Plot the Data¶

Let's look at our data in its raw form before doing any filtering.

In [3]:
trace1 = go.Scatter(
x=range(len(list(data['10 Min Std Dev']))),
y=list(data['10 Min Std Dev']),
mode='lines',
name='Wind Data'
)

layout = go.Layout(
showlegend=True
)

trace_data = [trace1]
fig = go.Figure(data=trace_data, layout=layout)
py.iplot(fig, filename='wind-raw-data-plot')

Out[3]:

#### Low-Pass Filter¶

A Low-Pass Filter is used to remove the higher frequencies in a signal of data.

fc is the cutoff frequency as a fraction of the sampling rate, and b is the transition band also as a function of the sampling rate. N must be an odd number in our calculation as well.

In [4]:
fc = 0.1
b = 0.08
N = int(np.ceil((4 / b)))
if not N % 2: N += 1
n = np.arange(N)

sinc_func = np.sinc(2 * fc * (n - (N - 1) / 2.))
window = 0.42 - 0.5 * np.cos(2 * np.pi * n / (N - 1)) + 0.08 * np.cos(4 * np.pi * n / (N - 1))
sinc_func = sinc_func * window
sinc_func = sinc_func / np.sum(sinc_func)

s = list(data['10 Min Std Dev'])
new_signal = np.convolve(s, sinc_func)

trace1 = go.Scatter(
x=range(len(new_signal)),
y=new_signal,
mode='lines',
name='Low-Pass Filter',
marker=dict(
color='#C54C82'
)
)

layout = go.Layout(
title='Low-Pass Filter',
showlegend=True
)

trace_data = [trace1]
fig = go.Figure(data=trace_data, layout=layout)
py.iplot(fig, filename='fft-low-pass-filter')

Out[4]:

#### High-Pass Filter¶

Similarly a High-Pass Filter will remove the lower frequencies from a signal of data.

Again, fc is the cutoff frequency as a fraction of the sampling rate, and b is the transition band also as a function of the sampling rate. N must be an odd number.

Only by performing a spectral inversion afterwards after setting up our Low-Pass Filter will we get the High-Pass Filter.

In [5]:
fc = 0.1
b = 0.08
N = int(np.ceil((4 / b)))
if not N % 2: N += 1
n = np.arange(N)

sinc_func = np.sinc(2 * fc * (n - (N - 1) / 2.))
window = np.blackman(N)
sinc_func = sinc_func * window
sinc_func = sinc_func / np.sum(sinc_func)

# reverse function
sinc_func = -sinc_func
sinc_func[(N - 1) / 2] += 1

s = list(data['10 Min Std Dev'])
new_signal = np.convolve(s, sinc_func)

trace1 = go.Scatter(
x=range(len(new_signal)),
y=new_signal,
mode='lines',
name='High-Pass Filter',
marker=dict(
color='#424242'
)
)

layout = go.Layout(
title='High-Pass Filter',
showlegend=True
)

trace_data = [trace1]
fig = go.Figure(data=trace_data, layout=layout)
py.iplot(fig, filename='fft-high-pass-filter')

Out[5]:

#### Band-Pass Filter¶

The Band-Pass Filter will allow you to reduce the frequencies outside of a defined range of frequencies. We can think of it as low-passing and high-passing at the same time.

In the example below, fL and fH are the low and high cutoff frequencies respectively as a fraction of the sampling rate.

In [6]:
fL = 0.1
fH = 0.3
b = 0.08
N = int(np.ceil((4 / b)))
if not N % 2: N += 1  # Make sure that N is odd.
n = np.arange(N)

# low-pass filter
hlpf = np.sinc(2 * fH * (n - (N - 1) / 2.))
hlpf *= np.blackman(N)
hlpf = hlpf / np.sum(hlpf)

# high-pass filter
hhpf = np.sinc(2 * fL * (n - (N - 1) / 2.))
hhpf *= np.blackman(N)
hhpf = hhpf / np.sum(hhpf)
hhpf = -hhpf
hhpf[(N - 1) / 2] += 1

h = np.convolve(hlpf, hhpf)
s = list(data['10 Min Std Dev'])
new_signal = np.convolve(s, h)

trace1 = go.Scatter(
x=range(len(new_signal)),
y=new_signal,
mode='lines',
name='Band-Pass Filter',
marker=dict(
color='#BB47BE'
)
)

layout = go.Layout(
title='Band-Pass Filter',
showlegend=True
)

trace_data = [trace1]
fig = go.Figure(data=trace_data, layout=layout)
py.iplot(fig, filename='fft-band-pass-filter')

Out[6]:
Still need help?