Scribd
Upload
2 views6 pages

CV Lab2

This document provides lab instructions for an undergraduate course on computer vision, focusing on image processing and manipulation using OpenCV and Matplotlib. It covers reading, displaying, and saving grayscale and color images, as well as manipulating image channels and blending images using various methods. Additionally, it includes tasks for students to apply their knowledge in practical coding exercises.

Uploaded by

morteza karimi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
2 views6 pages

CV Lab2

This document provides lab instructions for an undergraduate course on computer vision, focusing on image processing and manipulation using OpenCV and Matplotlib. It covers reading, displaying, and saving grayscale and color images, as well as manipulating image channels and blending images using various methods. Additionally, it includes tasks for students to apply their knowledge in practical coding exercises.

Uploaded by

morteza karimi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Fundamentals of Computer Vision (Undergrad) - B.

Nasihatkon
Spring 2025

Lab Instructions - session 2


Image’s processing and manipulation

Reading and displaying a grayscale image:


>>> import numpy as np
>>> import cv2
>>> I = cv2.imread('lake.jpg', cv2.IMREAD_GRAYSCALE)
>>> I.shape
>>> I.dtype
>>> I[10,20]
>>> cv2.imshow('Lake', I)
>>> cv2.waitKey(5000)
>>> cv2.destroyAllWindows()
>>> color_image = cv2.imread('lake.jpg')
>>> gray_I = cv2.cvtColor(color_image, cv2.COLOR_BGR2GRAY)
>>> gray_I.shape
>>> gray_I.dtype
>>> cv2.waitKey(5000)
>>> cv2.destroyAllWindows()
>>> cv2.imshow('Grayscale (Direct)',I)
>>> cv2.imshow('Grayscale (Converted)', gray_I)
>>> cv2.waitKey(0)
>>> cv2.destroyAllWindows()

Reading a color image:


>>> I = cv2.imread('lake.jpg', cv2.IMREAD_UNCHANGED)
>>> I.shape
>>> I.dtype
>>> I = cv2.imread('lake.jpg')
>>> I.shape
>>> I.dtype
>>> I[10,20]
>>> cv2.imshow('lake.jpg', I)
>>> cv2.waitKey()
>>> cv2.destroyAllWindows()

Side note: if we use cv2.waitKey() without waiting time, it will wait for the user to
press a key to continue.
Fundamentals of Computer Vision (Undergrad) - B. Nasihatkon
Spring 2025

Display the image using matplotlib:


>>> from matplotlib import pyplot as plt
>>> plt.imshow(I)
>>> plt.show()

●​ Are the image colors strange? Remember scipy/matplotlib use the RGB
system, while opencv uses the BGR system. Reading an image with opencv
and displaying with matplotlib makes the blue and red channels swapped.

>>> plt.imshow(I[:,:,::-1])
>>> plt.show()

Remember, I[:,:,::-1] reverses the order of the last dimension (the channels)
and hence converts BGR (opencv) to RGB (for displaying with matplotlib).
Could you suggest another way using cv2 for the above conversion?

Set the green channel equal to zero


remember openCV uses BGR: Blue: 0, Green: 1, Red: 2

>>> I[:,:,1] = 0​
>>> plt.imshow(I[:,:,::-1])​
>>> plt.show()

Saving an image
>>> cv2.imwrite('lake_nogreen.jpg', I)

Displaying ‘red’, ‘green’ and ‘blue’ channels


Open a code editor and write the following code (or open the file brg1.py in the
instructions folder). It displays the original image, the blue channel, the green
channel and the red channel of the image when the keys ‘o’, ‘b’, ‘g’ and ‘r’ are
pressed respectively. Notice that the green channel is brighter in green areas. So are
the red and blue channels.
import cv2
import numpy as np

I = cv2.imread('lake.jpg')
Fundamentals of Computer Vision (Undergrad) - B. Nasihatkon
Spring 2025

# notice that OpenCV uses BGR instead of RGB!


B = I[:,:,0]
G = I[:,:,1]
R = I[:,:,2]

cv2.imshow('win1',I)

while 1:
k = cv2.waitKey()

if k == ord('o'):
cv2.imshow('win1',I)
elif k == ord('b'):
cv2.imshow('win1',B)
elif k == ord('g'):
cv2.imshow('win1',G)
elif k == ord('r'):
cv2.imshow('win1',R)
elif k == ord('q'):
cv2.destroyAllWindows()
break

Task 1:
The above code shows the B, G and R channels as grayscale images. Change the
above code so that the blue channel is displayed in the blue, green channel in the
green, and red channel in red. Notice that since these are color images they have to
be mxnx3 arrays. For a purely red image, the green and blue channels are entirely
zero. The following code might help:​

B = np.zeros(I.shape, dtype=np.uint8)
print(B.shape)
B[:,:,0] = I[:,:,0]

Alternatively, you can create a copy of an image and change it:


B = I.copy()
# now change B so that it only contains the blue channel
Fundamentals of Computer Vision (Undergrad) - B. Nasihatkon
Spring 2025

Blending two images

Method 1:
import cv2
import numpy as np

I = cv2.imread('damavand.jpg')
J = cv2.imread('eram.jpg')

print(I.shape)
print(J.shape)

K = I.copy()
K[::2,::2,:] = J[::2,::2,:]

cv2.imshow("Image 1", I)
cv2.imshow("Image 2", J)
cv2.imshow("Blending", K)

cv2.waitKey(3000)
cv2.destroyAllWindows()

●​ What does the above piece of code do?


●​ What does K[::2,::2,:] = J[::2,::2,:] do? (notice that
J[::2,::2,:] represents every second row and every second column of J)
●​ Why have we written K = I.copy()instead of K = I?

Method 2
K = I//2+J//2

●​ What is the difference between methods 1 and 2?

numpy addition vs opencv addition


>>> a = np.array([1,2,3,4,5,6,7,8,9,10], dtype=np.uint8)
>>> a
>>> b = np.full((10,), 250, dtype=np.uint8)
>>> b
Fundamentals of Computer Vision (Undergrad) - B. Nasihatkon
Spring 2025

>>> a+b
>>> cv2.add(a,b)

●​ What is the difference between a+b and cv2.add(a,b)? Which one do you
think should be used for adding up images?

‌Blending with arbitrary ratios


0.8 of image I plus .2 of image J.
K = np.clip((0.8*I + 0.2*J), 0, 255).astype(np.uint8)

●​ Why have we used .astype(np.uint8)? print((0.8*I).dtype) to see why.

K = cv2.addWeighted(I,0.8,J,0.2, 0)
K = cv2.addWeighted(I,0.1,J,0.9, 0)
K = cv2.addWeighted(I,0.3,J,0.7, 0)

For floating point variables alpha, beta and gamma, running


cv2.addWeighted(I,alpha,J,beta,gamma)
is similar to running
np.clip((alpha*I + beta*J + gamma), 0, 255).astype(np.uint8)

Task 2:
Make an animated smooth transition from the image I to the J. You can use
“cv2.waitKey(n)” to delay for n milliseconds.

Task 3:
Create a smooth animation that changes image I from grayscale to color.
Hint: Grayscale images have equal values in all three color channels.
Fundamentals of Computer Vision (Undergrad) - B. Nasihatkon
Spring 2025

References
●​ https://docs.opencv.org/3.0-beta/doc/py_tutorials/py_core/py_table_of_conten
ts_core/py_table_of_contents_core.html#py-table-of-content-core
●​ https://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_gui/py_
image_display/py_image_display.html

You might also like