Computer Vision Tutorial — Lesson 2

Rakesh TS
10 min readDec 21, 2020

--

Note from author :

This tutorial is the foundation of computer vision delivered as “Lesson 2” of the series, there are more Lessons upcoming which would talk to the extend of building your own deep learning based computer vision projects. You can find the complete syllabus and table of content here

Target Audience : Final year College Students, New to Data Science Career, IT employees who wants to switch to data science Career .

Takeaway : Main takeaway from this article :

  1. Loading an Image from Disk
  2. Obtaining the ‘Height’, ‘Width’ and ‘Depth’ of Image
  3. Finding R,G,B components of the Image
  4. Drawing using OpenCV

Loading an Image from Disk:

Before we perform any operations or manipulations of an image, it is important for us to load an image of our choice to the disk. We will perform this activity using OpenCV. There are two ways we can perform this loading operations. One way is to load the image by simply passing the image path and image file to the OpenCV’s “imread” function. The other way is to pass the image through a command line argument using python module argparse.

Fig 2.1 Loading an Image from Disk by hard coding the image path and name in code

#Loading Image from disk

import cv2
image = cv2.imread(“C:/Sample_program/example.jpg”)
cv2.imshow(‘Image’, image)
cv2.waitKey(0)

Let’s create a file name Loading_image_from_disk.py in a notepad++. First we import our OpenCV library and contains our image processing functions. we import the library using the first line of code as cv2. The second line of code is where we read our image using cv2.imread function in OpenCV and we pass on the path of image as parameter, the path should also contain the file name with its image format extension .jpg , .jpeg , .png or .tiff .

syntax // image=cv2.imread(“path/to/your/image.jpg”) //

Absolute care has to be taken while specifying the file extension name. we are likely to receive the below error if we provide the wrong extension name

ERROR :

c:\Sample_program>python Loading_image_from_disk.py
Traceback (most recent call last):
File “Loading_image_from_disk.py”, line 4, in <module>
cv2.imshow(‘Image’, image)
cv2.error: OpenCV(4.3.0) C:\projects\opencv-python\opencv\modules\highgui\src\window.cpp:376: error: (-215:Assertion failed) size.width>0 && size.height>0 in function ‘cv::imshow’

The third line of code is where we actually display our image loaded. First parameter is a string, or the “name” of our window. The second parameter is the object to which image was loaded.

Finally, a call to cv2.waitKey pauses the execution of the script until we press a key on our keyboard. Using a parameter of “0” indicates that any keypress will un-pause the execution. Please feel free to run you program without having the last line of code in your program to see the difference.

Fig 2.2 Loading an Image using Argparse module

#Reading Image from disk using Argparse

import cv2
import argparse

apr = argparse.ArgumentParser()
apr.add_argument(“-i”, “ — image”, required=True, help=”Path to the image”)
args = vars(apr.parse_args())

image = cv2.imread(args[“image”])
cv2.imshow(‘Image’, image)
cv2.waitKey(0)

Knowing to read an image or file using command line argument (argparse) is an absolutely necessary skill to know, you can read more about command line argument in the link by clicking here → Click

First 2 lines are code are to import necessary libraries, here we import OpenCV and Argparse . We will be repeating this throughout the course.

Next 3 lines of code handle parsing the command line arguments. The only argument we need is — image : the path to our image on disk. Finally, we parse the arguments and store them in a dictionary called args .

Let’s take a second and quickly discuss exactly what the — image switch is. The — image “switch” (“switch” is a synonym for “command line argument” and the terms can be used interchangeably) is a string that we specify at the command line. This switch tells our Loading_image_from_disk.py script where the image we want to load lives on disk.

The last 3 lines are code are discussed earlier, cv2.imread function takes args[“image’] as a parameter, which is nothing but the image we provide in the command prompt. cv2.imshow displays the image , which is already stored in an image object from the previous line. The last line is pauses the execution of the script until we press a key on our keyboard.

One of the major advantage of using a argparse — command line argument is , we will be able to load images from different folder locations without having to change the image path in our program by dynamically passing the image path Ex — “ C:\CV_Material\image\sample.jpg “ in the command prompt as an argument while we execute our python program.

Fig 2.3 Loading Image fro different folder location using Argparse

c:\Sample_program>python Loading_image_from_disk.py — image C:\CV_Material\session1.JPG

Here, we are executing the Loading_image_from_disk.py python file from c:\sample_program location by passing “-image” parameter along with path of image C:\CV_Material\session1.JPG.

Obtaining the ‘Height’, ‘Width’ and ‘Depth’ of Image

Since images are represented as NumPy arrays, we can simply use the .shape attribute to examine the width, height, and the number of channels

By using the .shape attribute on the image object we just loaded. we can find the height,Width and Depth of the image. As discussed in the previous lesson — 1 , the Height and width of the image can be cross verified by opening the image in MS Paint. Refer the previous lesson. We will discuss about the depth of image in the upcoming lessons. Depth is also known as channel of an image. Colored images are usually of 3 channel because of the RGB composition in its pixels and grey scaled images are of 1 channel. This is something we had discussed in the previous Lesson-1

Fig 2.4 Prints the Height,Width and Depth of Image

#Obtaining Height,Width and Depth of an Image

import cv2
import argparse

apr = argparse.ArgumentParser()
apr.add_argument(“-i”, “ — image”, required=True, help=”Path to the image”)
args = vars(apr.parse_args())

# Only difference to the previous codes — shape attribute applied on image object

print(f’(Height,Width,Depth) of the image is: {image.shape}’)

image = cv2.imread(args[“image”])
cv2.imshow(‘Image’, image)
cv2.waitKey(0)

Output:

(Height,Width,Depth) of the image is: (538, 723, 3)

Only difference to the previous codes is the print statement that applies shape attribute on the loaded image object. f’ is the F-string @ formatted string that takes variables dynamically and prints.

f’ write anything here that you wan to see in the print statement : {variables,variables,object,object.attribute,}’

here we have used {object.attribute} inside the flower bracket for the .shape attribute to compute the height,width and depth of image object.

#Obtaining Height,Width and Depth separately

import cv2
import argparse

apr = argparse.ArgumentParser()
apr.add_argument(“-i”, “ — image”, required=True, help=”Path to the image”)
args = vars(apr.parse_args())

image = cv2.imread(args[“image”])

# NumPy array slicing to obtain the height, width and depth separately

print(“height: %d pixels” % (image.shape[0]))
print(“width: %d pixels” % (image.shape[1]))
print(“depth: %d” % (image.shape[2]))

cv2.imshow(‘Image’, image)
cv2.waitKey(0)

Output:

width: 723 pixels
height: 538 pixels
depth: 3

Here, instead of obtaining the (height,width,depth) together as tuple. we perform array slicing and obtain the height,width and depth of image individually. The 0th index of array contains the height of the image, 1st index contains the width of the image and 2nd index contains the depth of the image.

Finding R,G,B components of the Image

Fig 2.5 BGR value printed after taking the pixel co-ordinate position (y,x)

#Finding R,B,G of the Image at (x,y) position

import cv2
import argparse

apr = argparse.ArgumentParser()
apr.add_argument(“-i”, “ — image”, required=True, help=”Path to the image”)
args = vars(apr.parse_args())

image = cv2.imread(args[“image”])

# Receive the pixel co-ordinate value as [y,x] from the user, for which the RGB values has to be computed
[y,x] = list(int(x.strip()) for x in input().split(‘,’))

# Extract the (Blue,green,red) values of the received pixel co-ordinate
(b,g,r) = image[y,x]
print(f’The Blue Green Red component value of the image at position {(y,x)} is: {(b,g,r)}’)

cv2.imshow(‘Image’, image)
cv2.waitKey(0)

Output:

The Blue Green Red component value of the image at position (321, 308) is: (238, 242, 253)

Notice how the y value is passed in before the x value — this syntax may feel counter-intuitive at first, but it is consistent with how we access values in a matrix: first we specify the row number then the column number. From there, we are given a tuple representing the Blue, Green, and Red components of the image.

We can also change the color of a pixel at a given position by reversing operation

(b,g,r) = image[y,x] to image[y,x] = (b,g,r)

Here we assign the color in (BGR) to the image pixel co-ordinate . Lets give it a try by assigning RED color to the pixel at position (321,308) and validate the same by printing the pixel BGR at the given position

#Reversing the Operation to assign RGB value to the pixel of our choice

import cv2
import argparse

apr = argparse.ArgumentParser()
apr.add_argument(“-i”, “ — image”, required=True, help=”Path to the image”)
args = vars(apr.parse_args())

image = cv2.imread(args[“image”])

# Receive the pixel co-ordinate value as [y,x] from the user, for which the RGB values has to be computed
[y,x] = list(int(x.strip()) for x in input().split(‘,’))

# Extract the (Blue,green,red) values of the received pixel co-ordinate
image[y,x] = (0,0,255)
(b,g,r) = image[y,x]
print(f’The Blue Green Red component value of the image at position {(y,x)} is: {(b,g,r)}’)

cv2.imshow(‘Image’, image)
cv2.waitKey(0)

Output:

The Blue Green Red component value of the image at position (321, 308) is: (0, 0, 255)

In the above code, we receive the pixel co-ordinate through the command prompt by entering the value as shown in the below Fig 2.6 and assign the pixel co-ordinate red color by assigning (0,0,255) ie,(Blue,Green,Red) and validate the same by printing the input pixel co-ordinate.

Drawing using OpenCV

Lets learn how to draw different shapes like rectangle, square and circle using OpenCV by drawing circles to mask my eyes , rectangle to mask my lips and rectangle to mask the mantra-ray fish next to me.

The output should look like this:

Fig 2.6 Masked Photo of me with the mantra-ray fish.

This photo is masked with shapes using MS Paint, we will try to do the same using OpenCV by drawing circles around my eyes and rectangle to mask my lips and the matra-ray fish next to me.

We use cv2.rectangle method to draw rectangle and cv2.circle method to draw circle in OpenCV.

cv2.rectangle(image, (x1, y1), (x2, y2), (Blue, Green, Red), Thickness)

The cv2.rectangle method takes image as its first argument in which we want to draw our rectangle on. We want to draw on our image object that we have loaded, so we pass it into the method. The second argument is the starting (x1, y1) position of our rectangle — here we are starting our rectangle at point (156, 340). Then, we must provide an ending (x2, y2) point for the rectangle. We decide to end our rectangle at (360, 450). The next argument is the color of the rectangle we want to draw, here in this case we are passing black color in BGR format ie,(0,0,0). Finally the last argument we pass is the thickness of the line. we give -1 to draw solid shapes as seen in the Fig 2.6

Similarly we use cv2.circle method to draw circle

cv2.circle(image, (x, y), r, (Blue, Green, Red), Thickness)

The cv2.circle method takes image as its first argument in which we want to draw our rectangle on. We want to draw on our image object that we have loaded, so we pass it into the method. The second argument is the centre(x, y) position of our circle — here we have taken our circle at point (343, 243). The next argument is the radius of the circle we want to draw. The next argument is the color of the circle, here in this case we are passing RED color in BGR format ie,(0,0,255). Finally the last argument we pass is the thickness of the line. we give -1 to draw solid shapes as seen in the Fig 2.6

Ok ! By knowing all this. Lets try to accomplish what we started. Inorder for us to draw the shapes in the image, we need to identify the starting and ending (x,y) co-ordinates of the masking region to pass it to the respective method.

How ?

We will take the help of MS Paint for one more time. By placing the cursor over one of the co-ordinate (top left) or (bottom right) of the region to mask, the co-ordinates are shown in the highlighted portion of the MS Paint as shown in the Fig 2.7

Fig 2.7 MSPaint to find the co-ordinates of the masking region

Similarly we will take all the co-ordinate (x1,y1) (x2,y2) for all the masking regions as shown in the Fig 2.8

Fig 2.8 (x.y) Co-ordinates of masking regions

#Drawing using OpenCV to mask eyes,mouth and object near-by

import cv2
import argparse

apr = argparse.ArgumentParser()
apr.add_argument(“-i”, “ — image”, required=True, help=”Path to the image”)
args = vars(apr.parse_args())

image = cv2.imread(args[“image”])

cv2.rectangle(image, (415, 168), (471, 191), (0, 0, 255), -1)
cv2.circle(image, (425, 150), 15, (0, 0, 255), -1)
cv2.circle(image, (457, 154), 15, (0, 0, 255), -1)
cv2.rectangle(image, (156, 340), (360, 450), (0, 0, 0), -1)

# show the output image
cv2.imshow(“Output drawing “, image)
cv2.waitKey(0)

Result:

Fig 2.9 Desired Output

To read the other Lessons from this course, Jump to this article to find the complete syllabus and table of content

— — — — — — — — — — -> Click Here

--

--