Thresholding of Images

Aim

To segment the image using global thresholding, adaptive thresholding and Otsu's thresholding using python and OpenCV.

Software Required

Anaconda - Python 3.7
OpenCV

Algorithm

Step1:

Load the necessary packages.

Step2:

Read the Image and convert to grayscale.

Step3:

Use Global thresholding to segment the image.

Step4:

Use Adaptive thresholding to segment the image.

Step5:

Use Otsu's method to segment the image.

Step 6:

Display the results.

Program

Developed By: SRIJITH R

Reg No: 212221240054

Load the necessary packages

import cv2
import numpy as np
import matplotlib.pyplot as plt

Read the Image and convert to grayscale

ori_img=cv2.imread('kenny.jpg')
ori_img=cv2.resize(ori_img,(460,250))
gray_img=cv2.cvtColor(ori_img,cv2.COLOR_BGR2GRAY)

Use Global thresholding to segment the image

ret,thresh_img1=cv2.threshold(gray_img,86,255,cv2.THRESH_BINARY)
ret,thresh_img2=cv2.threshold(gray_img,86,255,cv2.THRESH_BINARY_INV)
ret,thresh_img3=cv2.threshold(gray_img,86,255,cv2.THRESH_TOZERO)
ret,thresh_img4=cv2.threshold(gray_img,86,255,cv2.THRESH_TOZERO_INV)
ret,thresh_img5=cv2.threshold(gray_img,100,255,cv2.THRESH_TRUNC)

Use Adaptive thresholding to segment the image

thresh_img6=cv2.adaptiveThreshold(gray_img,255,cv2.ADAPTIVE_THRESH_MEAN_C,cv2.THRESH_BINARY,11,2)
thresh_img7=cv2.adaptiveThreshold(gray_img,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,cv2.THRESH_BINARY,11,2)

Use Otsu's method to segment the image

ret,thresh_img8=cv2.threshold(gray_img,0,255,cv2.THRESH_BINARY+cv2.THRESH_OTSU)

Display the images

cv2.imshow('original',ori_img)
cv2.imshow('gray',gray_img)
cv2.imshow('binary threshold',thresh_img1)
cv2.imshow('binary to inverse threshold',thresh_img2)
cv2.imshow('to zero threshold',thresh_img3)
cv2.imshow('to zero to inverse threshold',thresh_img4)
cv2.imshow('truncate threshold',thresh_img5)
cv2.imshow('mean adaptive threshold',thresh_img6)
cv2.imshow('gaussian adaptive threshold',thresh_img7)
cv2.imshow('otsu thresold',thresh_img8)
cv2.waitKey(0)
cv2.destroyAllWindows()