Image Steganography using OpenCV in Python

Image Steganography is the process of hiding secret data in some image. In this post, we will hide one image inside another and convert it into another image and then extract back both the images from the previous image. The idea behind image-based Steganography is very simple. Images are composed of digital data (pixels), which describes what’s inside the picture, usually the colors of all the pixels. Since we know every image is made up of pixels and every pixel contains 3-values (red, green, blue). For example, suppose we have to hide img2 in img1, where both img1 and img2 are numpy nd array of pixel values. The size of img2 must be less than the size of img1. We are using color images, hence both will have 3 values (red, green, blue). Each pixel value varies from 0 to 255, so each pixel value is of 1 byte or 8 bits. Let img[i][j][l] be the pixel value at location (i, j) and of channel l where i varies from 0 to width and j varies from 0 to height and l varies from 0 to 2. Note: The quality of the new images is a little bit less than the old images.

Encoding

Let img1[i][j][l] and img2[i][j][l] be some pixel value of each image. Let v1 be 8 bits binary representation of img1[i][j][l] and v2 be 8 bits binary representation of img2[i][j][l]. Therefore, v3=v1[:4]+v2[:4], where, v3 is the first 4 bits of v1 and v2. Then we assign img1[i][j][l] to v3. Here img1 is the final image produced after encoding.

Decoding

Let img[i][j][l] be the pixel value of the image. Let v1 be 8 bits binary representation of img[i][j][l]. Let v2=v1[:4]+4 random bits and v3=v1[4:]+4 random bits. Then we assign img1[i][j][l] to v2 and img2[i][j][l] to v3. Here img1 and img2 are the final images produced after decoding. Implementation Lets' consider the images used are as follows: Image 1 Image 2 We want to hide image2 in image1. Below is the implementation. Python3

# Python program to demonstrate
# image steganography using OpenCV


import cv2
import numpy as np
import random


# Encryption function
def encrypt():
    
    # img1 and img2 are the
    # two input images
    img1 = cv2.imread('pic1.jpg')
    img2 = cv2.imread('pic2.jpg')
    
    for i in range(img2.shape[0]):
        for j in range(img2.shape[1]):
            for l in range(3):
                
                # v1 and v2 are 8-bit pixel values
                # of img1 and img2 respectively
                v1 = format(img1[i][j][l], '08b')
                v2 = format(img2[i][j][l], '08b')
                
                # Taking 4 MSBs of each image
                v3 = v1[:4] + v2[:4] 
                
                img1[i][j][l]= int(v3, 2)
                
    cv2.imwrite('pic3in2.png', img1)

    
# Decryption function
def decrypt():
    
    # Encrypted image
    img = cv2.imread('pic3in2.png') 
    width = img.shape[0]
    height = img.shape[1]
    
    # img1 and img2 are two blank images
    img1 = np.zeros((width, height, 3), np.uint8)
    img2 = np.zeros((width, height, 3), np.uint8)
    
    for i in range(width):
        for j in range(height):
            for l in range(3):
                v1 = format(img[i][j][l], '08b')
                v2 = v1[:4] + chr(random.randint(0, 1)+48) * 4
                v3 = v1[4:] + chr(random.randint(0, 1)+48) * 4
                
                # Appending data to img1 and img2
                img1[i][j][l]= int(v2, 2)
                img2[i][j][l]= int(v3, 2)
    
    # These are two images produced from
    # the encrypted image
    cv2.imwrite('pic2_re.png', img1)
    cv2.imwrite('pic3_re.png', img2)
    
    
# Driver's code
encrypt()
decrypt()

Output: After Encryption: After Decryption: Image 1 Image 2