Prerequisites: OpenCV Python Program to analyze an image using Histogram
Histogram of a digital image with intensity levels in the range of 0 to L-1 is a discrete function –
h(rk) = nk where rk = kth intensity value and no = number of pixels in the image with rk intensity value.
If the image has M rows and N columns, then the total number of pixels in the image is MN. The normalized histogram is obtained as follows:
p(rk) = nk/MN
Sometimes if the image is a dark, light, or low contrast, then the histogram of that image might not have pixels with minimum intensity (0) or maximum intensity(255) or both respectively. In such cases, the histogram of that image can be stretched by applying the following transformation.
Let a= maximum intensity level in the image Let b= minimum intensity level in the image Let rk= pixel value in the original image Let sk= pixel value in the stretched image constant= (255-0)/(a-b) Then sk= constant*rk
Algorithm for Histogram plotting
- Read the image
- Obtain the size of the image. Let m=rows and n=columns
- For all the intensity levels rk= 0 to 255
- Traverse all the rows ‘m’ and columns ‘n’
- Count the number of pixels for each intensity level
- Plot the graph of intensity values versus number of pixels in each intensity value
Algorithm to stretch the Histogram
- Find the minimum intensity value in the image say ‘a’ and maximum intensity value say ‘b’
- obtain a constant c= (255-0)/(a-b)
- Multiply this constant with each pixel in the image to obtain histogram stretching.
Observations It is observed that the dark input image improves after contrast stretching. Below is the implementation. Input Image: 
Python3
# import cv2, numpy, matplotlibimport cv2import numpy as npimport matplotlib.pyplot as plt # function to obtain histogram of an imagedef hist_plot(img): # empty list to store the count # of each intensity value count =[] # empty list to store intensity # value r = [] # loop to traverse each intensity # value for k in range(0, 256): r.append(k) count1 = 0 # loops to traverse each pixel in # the image for i in range(m): for j in range(n): if img[i, j]== k: count1+= 1 count.append(count1) return (r, count)img = cv2.imread('food.jpeg', 0)# To ascertain total numbers of rows and# columns of the image, size of the imagem, n = img.shaper1, count1 = hist_plot(img)# plotting the histogramplt.stem(r1, count1)plt.xlabel('intensity value')plt.ylabel('number of pixels')plt.title('Histogram of the original image') # Transformation to obtain stretchingconstant = (255-0)/(img.max()-img.min())img_stretch = img * constantr, count = hist_plot(img_stretch)# plotting the histogramplt.stem(r, count)plt.xlabel('intensity value')plt.ylabel('number of pixels')plt.title('Histogram of the stretched image') # Storing stretched Imagecv2.imwrite('Stretched Image 4.png', img_stretch) |
Output: 
Time Complexity: O(m)
Auxiliary Space: O(k)
