How to Use Canny Edge Detection in Computer Vision

Use cv2.Canny() from OpenCV to detect edges by specifying the input image and two threshold values. It finds strong and weak edges and connects them to highlight object boundaries clearly.

Syntax

The cv2.Canny() function detects edges in an image using two threshold values. The syntax is:

• image: The input grayscale image.
• threshold1: Lower boundary for edge detection.
• threshold2: Upper boundary for edge detection.
• apertureSize (optional): Size of the Sobel kernel (default is 3).
• L2gradient (optional): Boolean to use a more accurate gradient calculation.

python

edges = cv2.Canny(image, threshold1, threshold2, apertureSize=3, L2gradient=False)

Example

This example loads a grayscale image, applies Canny edge detection, and displays the original and edge images side by side.

python

import cv2
import numpy as np

# Load image in grayscale
image = cv2.imread('input.jpg', cv2.IMREAD_GRAYSCALE)

# Check if image loaded
if image is None:
    raise FileNotFoundError('Image file not found.')

# Apply Canny edge detection
edges = cv2.Canny(image, threshold1=100, threshold2=200)

# Stack images horizontally for comparison
combined = np.hstack((image, edges))

# Show the images
cv2.imshow('Original and Edges', combined)
cv2.waitKey(0)
cv2.destroyAllWindows()

Output

A window opens showing the original grayscale image on the left and the detected edges on the right.

Common Pitfalls

• Wrong image format: Canny requires a grayscale image, not color.
• Threshold values: Setting thresholds too low or too high can miss edges or detect noise.
• Ignoring noise: Applying Canny on noisy images without smoothing can cause false edges.

Always convert to grayscale and consider applying Gaussian blur before Canny.

python

import cv2

# Wrong way: Using color image
image_color = cv2.imread('input.jpg')
edges_wrong = cv2.Canny(image_color, 100, 200)  # May give incorrect results

# Right way: Convert to grayscale and blur
image_gray = cv2.cvtColor(image_color, cv2.COLOR_BGR2GRAY)
image_blur = cv2.GaussianBlur(image_gray, (5, 5), 1.4)
edges_right = cv2.Canny(image_blur, 100, 200)

Quick Reference

Parameter	Description	Typical Values
image	Input grayscale image	Grayscale image array
threshold1	Lower threshold for edge detection	50-150
threshold2	Upper threshold for edge detection	100-300
apertureSize	Sobel kernel size for gradient	3 (default), 5, or 7
L2gradient	Use precise gradient calculation	False (default) or True

Key Takeaways

Always convert your image to grayscale before applying Canny edge detection.

Choose threshold values carefully to balance edge detection and noise suppression.

Apply Gaussian blur to reduce noise and improve edge detection quality.

Use cv2.Canny() with proper parameters to detect clear edges.

Visualize results side by side to verify edge detection effectiveness.