How to Apply Threshold in OpenCV for Computer Vision

Use OpenCV's cv2.threshold() function to convert grayscale images into binary images by setting a pixel intensity cutoff. This function takes the source image, threshold value, max value, and thresholding type as inputs and outputs the thresholded image and the used threshold value.

Syntax

The cv2.threshold() function syntax is:

• src: Grayscale input image.
• thresh: Threshold value to classify pixel intensities.
• maxval: Value assigned to pixels exceeding the threshold.
• type: Thresholding type (e.g., binary, binary inverse).

The function returns a tuple: the used threshold and the thresholded image.

python

retval, dst = cv2.threshold(src, thresh, maxval, type)

Example

This example loads a grayscale image, applies a binary threshold at 127, and shows the original and thresholded images.

python

import cv2

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

# Apply binary threshold
retval, thresh_img = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY)

# Save thresholded image
cv2.imwrite('thresholded.jpg', thresh_img)

print(f'Threshold used: {retval}')

Output

Threshold used: 127.0

Common Pitfalls

Common mistakes include:

• Using a color image instead of grayscale, which causes errors or unexpected results.
• Choosing an inappropriate threshold value, leading to poor segmentation.
• Confusing threshold types, such as using THRESH_BINARY_INV when THRESH_BINARY is intended.

Always convert images to grayscale before thresholding and experiment with threshold values for best results.

python

import cv2

# Wrong: Using color image
img_color = cv2.imread('input.jpg')
retval, thresh_wrong = cv2.threshold(img_color, 127, 255, cv2.THRESH_BINARY)  # This will fail or give wrong output

# Right: Convert to grayscale first
img_gray = cv2.cvtColor(img_color, cv2.COLOR_BGR2GRAY)
retval, thresh_right = cv2.threshold(img_gray, 127, 255, cv2.THRESH_BINARY)

Quick Reference

Threshold Type	Description
cv2.THRESH_BINARY	Pixel > thresh set to maxval, else 0
cv2.THRESH_BINARY_INV	Pixel > thresh set to 0, else maxval
cv2.THRESH_TRUNC	Pixel > thresh set to thresh, else unchanged
cv2.THRESH_TOZERO	Pixel > thresh unchanged, else 0
cv2.THRESH_TOZERO_INV	Pixel > thresh set to 0, else unchanged

Key Takeaways

Always convert images to grayscale before applying threshold in OpenCV.

Choose the threshold value carefully to get meaningful segmentation results.

Use the correct threshold type to match your desired output.

The cv2.threshold function returns both the threshold used and the thresholded image.

Experiment with different threshold types for different image processing tasks.