How to perform 5 most famous types of Thresholding Techniques in Python using OpenCV

3 min readJan 30, 2022

In today’s blog, we are going to perform one of the most important operations of image processing that is thresholding. So without any further due, let’s dive into it.

Read the full article with source code here — https://machinelearningprojects.net/thresholding/

Let’s do it…

Step 1 — Import the libraries required for thresholding.

import cv2
import matplotlib.pyplot as plt

Step 2 — Let's read the grayscale image.

img = cv2.imread(‘gray21.512.tiff’)

Step 3 — Lets instantiate some values.

th = 127
max_val = 255

Here we have set our threshold value to 127.
Also, we have set our max value to 255.

Step 4 — Let's perform the thresholding operation using the following 5 methods.

ret, o1 = cv2.threshold(img, th, max_val, cv2.THRESH_BINARY)
ret, o2 = cv2.threshold(img, th, max_val, cv2.THRESH_BINARY_INV)
ret, o3 = cv2.threshold(img, th, max_val, cv2.THRESH_TOZERO)
ret, o4 = cv2.threshold(img, th, max_val, cv2.THRESH_TOZERO_INV)
ret, o5 = cv2.threshold(img, th, max_val, cv2.THRESH_TRUNC)

cv2.threshold() returns 2 values, first is the return value(True or False) and second is the output image.
cv2.THRESH_BINARY method gives max value as soon as value crosses threshold.
cv2.THRESH_BINARY_INV method gives max value till it does not crosses threshold.
cv2.THRESH_TOZERO method gives 0 value till it does not crosses threshold.
cv2.THRESH_TOZERO_INV method gives 0 as it crosses threshold.
cv2.THRESH_TRUNC method gives threshold value as it crosses threshold.

NOTE — Refer to the resulting image below for further understanding.

Step 5 — Plot the results.

output = [img, o1, o2, o3, o4, o5]

titles = ['Original', 'Binary', 'Binary Inv', 'Zero', 'Zero Inv', 'Trunc']

for i in range(6):
    plt.subplot(2, 3, i + 1)
    plt.imshow(output[i])
    plt.title(titles[i])
    plt.xticks([])
    plt.yticks([])

plt.show()

Results

NOTE — Thresholding can be further optimized by using OTSU Thresholding.

ret, o1 = cv2.threshold(img, 0, max_val, cv2.THRESH_BINARY+cv2.THRESH_OTSU)
ret, o2 = cv2.threshold(img, 0, max_val, cv2.THRESH_BINARY_INV+cv2.THRESH_OTSU)
ret, o3 = cv2.threshold(img, 0, max_val, cv2.THRESH_TOZERO+cv2.THRESH_OTSU)
ret, o4 = cv2.threshold(img, 0, max_val, cv2.THRESH_TOZERO_INV+cv2.THRESH_OTSU)
ret, o5 = cv2.threshold(img, 0, max_val, cv2.THRESH_TRUNC+cv2.THRESH_OTSU)

Simply just add a cv2.THRESH_OTSU as a flag as shown above to implement it.
The main advantage that we have with OTSU is that in this method we don’t need to specify the threshold value (just specify 0 there) like in previous methods(without the OTSU method), it determines it automatically.

NOTE — Read more about thresholding operation here.

Do let me know if there’s any query regarding this topic by contacting me on email or LinkedIn.

For further code explanation and source code visit here — https://machinelearningprojects.net/thresholding/

So this is all for this blog folks, thanks for reading it and I hope you are taking something with you after reading this and till the next time 👋…

Read my previous post: HOW TO SPLIT AND MERGE CHANNELS IN PYTHON USING OPENCV

Check out my other machine learning projects, deep learning projects, computer vision projects, NLP projects, Flask projects at machinelearningprojects.net.