Session 08🔍
Region Labeling & Edge Detection
Colorizing segmented regions and detecting edges with Prewitt gradient operators.
Learning Objectives
- ✓Map labeled regions to custom colors
- ✓Compute horizontal and vertical Prewitt gradients
- ✓Combine gradient components into edge magnitude
- ✓Threshold edges at multiple levels
Color Mapping by Region
After labeling, each region can be assigned a distinct color by creating a colormap indexed by label number. This makes it easy to visually inspect segmentation results.
Ex8_1 — Colorize labeled regionsPython
import cv2
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.colors as mcolors
img = cv2.cvtColor(cv2.imread('objects.jpg'), cv2.COLOR_BGR2RGB)
gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
_, binary = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
num_labels, labels = cv2.connectedComponents(binary)
# Create a colormap for N labels
cmap = plt.get_cmap('tab20', num_labels)
colored = cmap(labels / num_labels) # RGBA float array
colored_uint8 = (colored[:, :, :3] * 255).astype(np.uint8)Prewitt Operators
Prewitt operators compute image gradients in horizontal and vertical directions. The kernels approximate the first derivative. Gx detects vertical edges; Gy detects horizontal edges.
Ex8_2 — Prewitt edge detectionPython
import cv2
import numpy as np
from scipy.signal import convolve2d
img = cv2.imread('scene.jpg', cv2.IMREAD_GRAYSCALE).astype(np.float64)
# Prewitt kernels
Gx_kernel = np.array([[-1, 0, 1],
[-1, 0, 1],
[-1, 0, 1]], dtype=np.float64)
Gy_kernel = np.array([[-1,-1,-1],
[ 0, 0, 0],
[ 1, 1, 1]], dtype=np.float64)
Gx = convolve2d(img, Gx_kernel, mode='same', boundary='symm')
Gy = convolve2d(img, Gy_kernel, mode='same', boundary='symm')
# Gradient magnitude
G = np.sqrt(Gx**2 + Gy**2)Edge Thresholding
After computing the gradient magnitude, apply different thresholds to control how many edges are visible. Higher threshold = only strong edges. Lower threshold = more edges including noise.
Ex8_2 — Threshold at multiple levelsPython
G_norm = (G / G.max() * 255).astype(np.uint8)
thresholds = [30, 60, 100]
results = []
for T in thresholds:
_, edge_map = cv2.threshold(G_norm, T, 255, cv2.THRESH_BINARY)
results.append(edge_map)