Signal and Image Processing Modules

SciPy includes powerful modules for signal and image processing, enabling you to:

• Filter time-series or spatial data
• Perform frequency analysis
• Apply transformations and basic image manipulation

The two main modules are:

• scipy.signal – Works with 1D and 2D signals (e.g., audio, time series, images)
• scipy.ndimage – Offers advanced image processing functions

These tools are widely used in engineering, data analysis, and scientific research.

Copy
import numpy as np
import matplotlib.pyplot as plt
from scipy import signal, misc

# Sampling parameters
fs = 500.0  # Hz
t = np.arange(0, 1.0, 1/fs)

# Create a noisy signal
sig = np.sin(2*np.pi*5*t) + 0.5*np.sin(2*np.pi*50*t)

# Design low-pass filter (cutoff 10 Hz)
b, a = signal.butter(N=4, Wn=10/(fs/2), btype='low')

# Apply filter
filtered = signal.filtfilt(b, a, sig)

# Plot
plt.plot(t, sig, label="Original", alpha=0.5)
plt.plot(t, filtered, label="Filtered", linewidth=2)
plt.xlabel("Time [s]")
plt.ylabel("Amplitude")
plt.title("Low-Pass Butterworth Filter")
plt.legend()
plt.show()

signal.butter() designs the filter, and signal.filtfilt() applies it without introducing a phase shift.

Copy
f, t_spec, Sxx = signal.spectrogram(sig, fs)
plt.pcolormesh(t_spec, f, Sxx, shading='gouraud')
plt.ylabel("Frequency [Hz]")
plt.xlabel("Time [s]")
plt.title("Spectrogram")
plt.show()

A spectrogram shows how the frequency content of a signal changes over time — useful for audio analysis and vibration monitoring.

Copy
# Load sample image
face = misc.face(gray=True)

# Create Gaussian blur kernel
kernel_size = 15
sigma = 3.0
x = np.linspace(-sigma, sigma, kernel_size)
gauss_kernel_1d = np.exp(-x**2 / (2 * sigma**2))
gauss_kernel_1d /= gauss_kernel_1d.sum()
gauss_kernel_2d = np.outer(gauss_kernel_1d, gauss_kernel_1d)

# Apply convolution
blurred_face = signal.convolve2d(face, gauss_kernel_2d, mode='same', boundary='symm')

# Plot
plt.subplot(1, 2, 1)
plt.imshow(face, cmap='gray')
plt.title("Original")
plt.axis('off')

plt.subplot(1, 2, 2)
plt.imshow(blurred_face, cmap='gray')
plt.title("Blurred")
plt.axis('off')

plt.show()

signal.convolve2d() applies a 2D convolution — in this case, using a Gaussian kernel to smooth the image.

In the next lesson, we’ll conclude Chapter 3 with the Final Quiz on SciPy Fundamentals to test your knowledge from all previous lessons.