SciPy — Image Processing

Read and display

from scipy import ndimage
import numpy as np

# Create test image
image = np.random.rand(100, 100)

Filtering

from scipy.ndimage import gaussian_filter, uniform_filter

# Gaussian blur
blurred = gaussian_filter(image, sigma=1)

# Mean filter
filtered = uniform_filter(image, size=3)

Edge detection

from scipy.ndimage import sobel

# Compute gradients
sx = sobel(image, axis=0)
sy = sobel(image, axis=1)
edges = np.hypot(sx, sy)

Morphology

from scipy.ndimage import binary_dilation, binary_erosion

# Create binary image
binary = image > 0.5

# Dilate
dilated = binary_dilation(binary)

# Erode
eroded = binary_erosion(binary)

Transformations

from scipy.ndimage import rotate, shift, zoom

# Rotate
rotated = rotate(image, 45)

# Shift
shifted = shift(image, (10, 10))

# Zoom
zoomed = zoom(image, 2)

Labeling

from scipy.ndimage import label

# Label connected components
labeled, num_features = label(binary)
print(f"Found {num_features} objects")

Measurements

from scipy.ndimage import center_of_mass, find_objects

# Center of mass
com = center_of_mass(image)

# Find objects
objects = find_objects(labeled)

Mini Practice

1. Apply Gaussian blur
2. Detect edges
3. Perform morphological operations
4. Label connected components

Up Next

Continue with Sparse Matrices - Efficient matrices.