Image Warping and Morphing

Image Warping and Morphing are two methods used in computer graphics and image processing. They are used for tasks like transformation which include rotation, scaling etc., and even manipulation which is used to give creative effects. These methods are mostly used in video games, movies, face filters, and even for medical purposes.

In this article, we will go through these topics. We will be using Python’s OpenCV package to implement these techniques. I have used PyCharm IDE to perform these techniques.

Key Words

Before we dive into image warping and morphing, let’s first understand a few basic terminologies:

Image Warping

The image warping process means that it is a process of distorting an image using various transformations like scaling, rotation, translation, and more. This is also known as geometric transformation. Using this method we can see the image from another angle.

Example: If you have an image of a rectangle, and you want to rotate it by 45 degrees. You would create a rotation matrix, apply it to the image, and obtain its rotated version. This is an example of simple warping.

Morphing

Image morphing is a special form of image warping which is a simple and smooth transition between two or more images. In simple words, its like a transformation of one image to another. This is mostly seen in movies and animations.

For example: Suppose we want to transform one face into another face. So, the first step would be to select the corresponding features like eyes, nose, and mouth in both images. Then, we would create a smooth transition between these features to create a morphing effect. This is similar to an age filter.

Pixel

Pixel is the smallest unit of an image. Each pixel is a single point in the image and has attributes like color and intensity.

Transformation Matrix

Transformation matrix is used to represent geometric transformations like rotation, scaling, and translation. This helps us to control the image warping.

Implementation of Image Processing Techniques

Firstly, open your Pycharm IDE. After that, we need to install the opencv library into our environment. For that we go to the terminal of the IDE and type the following:

pip install opencv-python

installing opencv in pycharm terminal

This will install the opencv library and we are ready to use this package now.

Image Warping Process

Let us first dive into the image warping process step-by-step:

Step 1: Importing libraries

Now as we have our library installed, we have to import the libraries we are going to use. We are going to import open-cv as we are doing these tasks in python programming language. We would also need numpy library for numerical purposes. The code is as follows:

import cv2

import numpy as np

Step 2: Image Loading

We have to then load the image into the programming environment to work with it. The function used here is ‘cv2.imread’ function. The argument passed must be the path of the image present on your desktop.

image = cv2.imread(“pathofimage”)

Step 3: Choose your source and destination points

Now that we have our image we need to choose the source points and destination points, which means the points of image and where they should be after the transformation.

src_points = np.float32([[0,0],[100,0],[0,100],[100,100]])

dst_points = np.float32([[60, 60],[40,60],[0,40],[40,40]])

Here, the points are defined in array of numpy package. The order is top-left, top-right, bottom-left, bottom-right for the points in array.

Step 4: Calculating Transformation Matrix

Based on the chosen transformation, we need to construct a transformation matrix. For that we would use a pre-defined function called ‘getPerspectiveTransform‘.

matrix = cv2.getPerspectiveTransform(src_points, dst_points)

Step 5: Apply Transformation

Now using this matrix, we can apply transformation to the image.

warped_image = cv2.warpPerspective(image, matrix, (image.shape[1], image.shape[0]))

The ‘cv2.warpPerspective‘ function is used to apply the perspective warp to the image using the transformation ‘matrix‘. The image.shape[1] and image.shape[0] are the width and height of the image respectively. These are nothing but the dimensions of the image.

Step 6: Display Result

Now display the original and warped image to see the effects of the transformation using ‘imshow‘ function.

cv2.imshow(‘Original Image’, image)

cv2.imshow(‘Warped Image’, warped_image)

Step 7: Closing windows

We can press ‘0’ and close the image windows, which is done by the two functions given below, where first, the script waits till we press ‘0’ key on keyboard, and then closes the window when pressed. The functions used are ‘waitKey()‘ and ‘destroyAllWindows()‘.

cv2.waitKey(0)

cv2.destroyAllWindows()

Code

Python3

#import libraries
import cv2
import numpy as np

# load the image using file path
image = cv2.imread("C:/Users/preet/Desktop/warp.png")

# defining source and destination points
src_points = np.float32([[0, 0], [image.shape[1] - 1, 0], [0, image.shape[0] - 1], [image.shape[1] - 1, image.shape[0] - 1]])
dst_points = np.float32([[100, 100], [image.shape[1] - 100, 100], [0, image.shape[0] - 1], [image.shape[1] - 1, image.shape[0] - 1]])

# calculatint the perspective transformation matrix
matrix = cv2.getPerspectiveTransform(src_points, dst_points)

# apply the perspective warp to the image using dimensions
warped_image = cv2.warpPerspective(image, matrix, (image.shape[1], image.shape[0]))

# display the original and warped images
cv2.imshow('Original Image', image)
cv2.imshow('Warped Image', warped_image)

# wait for 0 to be pressed and then close the windows
cv2.waitKey(0)
cv2.destroyAllWindows()

Output

before warping

after warping

Image Morphing Process

Image morphing is an extension of image warping method, which is smooth transition between two images. Here we will see the steps needed for simple image morphing:

Step 1: Importing libraries

Similar to image warping, the first step is to import the libraries. We are going to import open-cv. The code is as follows:

import cv2

Step 2: Image Paths and Resizing

Now, we need to select two or more images we want to morph. These images need to have similar dimensions, so that the transition would appear smooth. Hence, we would resize the images. Here, ‘(x,x)‘ is the dimension of the image.

image_source = cv2.imread( “pathofimage1.jpg”)

image_source = cv2.resize(image_source, (x, x)

image_destination = cv2.imread(“pathofimage2.jpg”)

image_destination = cv2.resize(image_destination, (x, x))

Step 3: Image Blending

Now, we create a ‘for’ loop which will run for many steps, where each step is a step in the process of morphing. Now, we calculate the percentage to control the blending inside the for loop.

Here, ‘addWeighted‘ function helps in blending the source image and destination image using the percentage difference between the source image and destination image. In simple words, as the source images vanishes, the destination images becomes more clear. The morphed images gets displayed continuously until the for loop ends which is called the display of Morph Sequence.

for i in range(100):

percentage = float(i / 100)

imgAdd = cv2.addWeighted(image_source, 1 – percentage, image_destination, percentage, 0)

cv2.imshow(“Morphed”, imgAdd)

cv2.waitKey(100)

Step 4: Closing the window

This is final step to close the windows when 0 is pressed.

cv2.waitKey(0)

cv2.destroyAllWindows()

Now let’s see the how to do simple morphing using Python Open CV package:

Code

Python3

#import libraries
import cv2

#images paths in source and destination
source = "C:/Users/preet/Desktop/bird 1.png"
destination = "C:/Users/preet/Desktop/bird 2.png"

#reading images and resizing it
image_source = cv2.imread(source)
image_source = cv2.resize(image_source, (300, 300))

image_destination = cv2.imread(destination)
image_destination = cv2.resize(image_destination, (300, 300))

#morphing in 100 steps
for i in range(100):
    percentage = float(i / 100)
    print(percentage)
    imgAdd = cv2.addWeighted(image_source, 1 - percentage, image_destination, percentage, 0)
    cv2.imshow("Morphed", imgAdd)
    cv2.waitKey(100)

#waiting till 0 is pressed
cv2.waitKey(0)
#closing all windows
cv2.destroyAllWindows()

Output

bird morph

Conclusion

In conclusion, image warping and morphing are powerful techniques with many applications in computer graphics and image processing. Understanding these concepts and following the step-by-step processes will help us to apply these methods effectively in our projects and creative tasks.

FAQs on Image Warping and Morphing

1. What Is The Difference Between Image Warping and Morphing?

Image warping is a process of transforming an image using various methods like scaling, rotation, translation etc. Using this method we can see an image from another angle, whereas Morphing is special form of image warping that creates a smooth transition between images.

2. Can Image Warping Be Used For Image Enhancement?

Yes, image warping can be used for image enhancement like fixing the shape of image, changing the size etc.

3. What Software Or Libraries Can I Use For Image Warping And Morphing?

We can use OpenCV library, MATLAB software to do these image processing tasks. There are even more tools available which can be useful in performing these techniques.

4. Are There Any Real-World Applications Of Image Morphing?

Yes, image morphing is used in movies and animations for creating special effects. It is also sometimes used in medical imaging for visualizing some anatomical changes which we can’t see via naked eye. It can also be used to show how your face changes as you grow old.

5. Can Image Morphing Be Used For Video Sequences?

Yes, image morphing techniques can be even used to create video sequences by using morphing technique between frames, which creates smooth transitions in videos.