Skip to main content

Getting Started with OpenCV - Image Loading, Displaying, Saving and Dimensions


Hi Guys, Welcome! to step by step tutorial of Getting Started with OpenCV - Image Loading, Displaying, Saving and Dimensions. OpenCV is the most popular and open source python library in the field of image processing and Computer Vision. From a very long time I was thinking about to start a blog series about OpenCV, Image processing and Computer Vision and finally I am starting this series of OpenCV blogs with this first blog Getting Started with OpenCV. So, In this very first tutorial we will start with OpenCV and image processing basics. We are going to learn:

  1. Import the OpenCV Model in Python
  2. Load Images
  3. Display Images
  4. Save Images
  5. Getting the Image Dimensions

Prerequisites and Installation

We will start with Prerequisites and Installation for OpenCV Python library -

  1. You should have Windows or Linux or Mac OS installed on your PC.
  2. Make sure you should have internet up and running for the installations.
  3. Install any IDE of your choice (Like VS Code, Jupyter Lab/Notebook, Pycharm etc...). For OpenCV my personal choice is Jupyter Lab/Notebook.
  4. Install OpenCV Python library with pip - pip install opencv-python.

Once all the above mentioned requirements are fulfilled, we are good to go.

Loading the OpenCV Library

import cv2

We will start by importing cv2 module of OpenCV.

import cv2

  1. import Statement

    • The import keyword in Python is used to bring in external libraries or modules into the current script. This allows you to use functions, classes, and methods defined in that module without rewriting them.

  2. cv2 Module

    • cv2 is the module name for OpenCV (Open Source Computer Vision Library) in Python.
    • OpenCV is a powerful library used for image processing, computer vision, and machine learning tasks.

What Happens When This Line is Executed?

  • Python searches for the cv2 module in its installed packages.
  • If OpenCV is installed (pip install opencv-python or pip install opencv-python-headless for headless environments), Python loads the module.
  • After importing, you can use OpenCV functions like reading images (cv2.imread()), displaying images (cv2.imshow()), and performing various image-processing operations.
# Let's see what version we're running
print(cv2.__version__)

4.11.0

  1. print() Function

    • The print() function in Python is used to display output in the console.
    • Whatever is inside print() will be evaluated and displayed.

  2. cv2.__version__ Attribute

    • cv2.__version__ is a special attribute of the OpenCV module that returns the installed OpenCV version as a string.
    • It helps in verifying which version of OpenCV is installed in your Python environment.

What Happens When This Line is Executed?

  • Python retrieves the __version__ attribute from the cv2 module.
  • It prints the OpenCV version in the format "major.minor.patch" (e.g., "4.5.3" or "4.8.0").

Example Output:

If OpenCV 4.11.0 is installed, the output would be:

4.11.0

This is useful when debugging or ensuring compatibility with certain OpenCV features.

Loading Images

# Load an image using 'imread' specifying the path to image
image = cv2.imread('./image/fruits.webp')

  1. cv2.imread() Function

    • cv2.imread() is an OpenCV function used to read an image from a file.
    • It loads the image as a NumPy array, which can be processed using OpenCV functions.

  2. './image/fruits.webp' (File Path)

    • This is the relative file path of the image that we want to load.
    • './image/fruits.webp' means:
      • . → Current directory
      • image/ → A folder named "image" inside the current directory
      • fruits.webp → The image file name (a WebP format image)
    • If the file does not exist at the given path, cv2.imread() will return None.

  3. image = (Variable Assignment)

    • The image is stored in the image variable as a NumPy array.
    • Each pixel of the image is represented as an array of BGR (Blue, Green, Red) values.
    • If the image is grayscale, the array will have only one channel.

What Happens When This Line is Executed?

  • OpenCV searches for the fruits.webp file in the ./image/ directory.
  • If found, it loads the image into the image variable as a NumPy array.
  • If not found, image will be None.

Displaying Images

from matplotlib import pyplot as plt

#Show the image with matplotlib
plt.imshow(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
plt.show()

We will start with:

from matplotlib import pyplot as plt

plt.imshow(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
plt.show()

1. Importing Matplotlib's pyplot Module

from matplotlib import pyplot as plt
  • matplotlib is a Python library for data visualization.
  • The pyplot module (plt) provides a collection of functions to create and display plots, including images.

2. Displaying the Image with plt.imshow()

plt.imshow(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
  • plt.imshow() is used to display an image.
  • However, OpenCV loads images in BGR (Blue-Green-Red) format, while Matplotlib expects RGB (Red-Green-Blue) format.
  • To fix the color mismatch, we use:
    cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
    • cv2.cvtColor() converts the image from BGR to RGB format.

3. Displaying the Image with plt.show()

plt.show()
  • plt.show() renders and displays the image in a new window.

What Happens When This Code Runs?

  1. The image is loaded in BGR format using OpenCV (cv2.imread()).
  2. The cv2.cvtColor() function converts it to RGB format.
  3. plt.imshow() plots the image correctly.
  4. plt.show() displays the image.

Why Do We Need cv2.cvtColor(image, cv2.COLOR_BGR2RGB)?

  • OpenCV uses BGR format, whereas Matplotlib expects RGB.
  • Without conversion, the colors would appear incorrect (e.g., blue would appear as red, and vice versa).
def imshow(title = "", image = None):
    plt.imshow(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
    plt.title(title)
    plt.show()

# Let's test it out
imshow("Displaying Our First Image", image)

Saving Images

# Simply use 'imwrite' specificing the file name and the image to be saved
cv2.imwrite('./image/output.jpg', image)

# Or save it as a PNG (Portable Network Graphics) which is a Lossless bitmap image format
cv2.imwrite('./image/output.png', image)

True

cv2.imwrite('./image/output.jpg', image)
cv2.imwrite('./image/output.png', image)

1. cv2.imwrite() Function

  • cv2.imwrite() is an OpenCV function used to save an image to a specified file.
  • Syntax:
    cv2.imwrite(filename, image)
    • filename: The path where the image will be saved.
    • image: The NumPy array containing the image data.

2. Saving Image as JPG (output.jpg)

cv2.imwrite('./image/output.jpg', image)
  • Saves the image in JPEG format inside the image/ directory with the name output.jpg.
  • JPEG is a compressed format, meaning the image size will be smaller but may lose some quality.

3. Saving Image as PNG (output.png)

cv2.imwrite('./image/output.png', image)
  • Saves the same image in PNG format as output.png.
  • PNG is a lossless format, preserving the image quality but potentially having a larger file size.

What Happens When This Code Runs?

  1. OpenCV writes the image NumPy array to a file.
  2. The image gets saved in the specified format (.jpg or .png).
  3. If the directory ./image/ does not exist, an error might occur.
# Import numpy
import numpy as np

print(image.shape)

(1024, 1024, 3)

import numpy as np
print(image.shape)

1. Importing NumPy

import numpy as np
  • numpy is a powerful Python library for numerical computing.
  • OpenCV stores images as NumPy arrays, making it easy to manipulate images using NumPy functions.

2. Printing the Shape of the Image

print(image.shape)
  • image is assumed to be a NumPy array (loaded using cv2.imread()).
  • .shape is an attribute of a NumPy array that returns its dimensions in the form of:
    (height, width, channels)
    where:
    • height → Number of rows (pixels) in the image.
    • width → Number of columns (pixels) in the image.
    • channels → Number of color channels (typically 3 for RGB/BGR images, 1 for grayscale images).

Example Outputs

1. If the image is a color image (BGR)

print(image.shape)

Output:

(720, 1280, 3)

This means:

  • The image has 720 pixels in height.
  • The image has 1280 pixels in width.
  • The image has 3 channels (BGR color format).

2. If the image is grayscale

print(image.shape)

Output:

(720, 1280)

This means:

  • The image has 720 pixels in height.
  • The image has 1280 pixels in width.
  • There is no third value because the image is grayscale (single channel).

How to Handle Grayscale Images?

If you need to check whether the image is grayscale or color:

if len(image.shape) == 2:
    print("The image is grayscale.")
else:
    print("The image is color (BGR).")

# To access a dimension, simply index it by using 0, 1 or 2.
image.shape[0]

1024

1. image.shape Attribute

  • image is a NumPy array that stores the image data (loaded using cv2.imread()).
  • image.shape returns a tuple representing the dimensions of the image:
    (height, width, channels)
    • height → Number of rows (pixels) → image.shape[0]
    • width → Number of columns (pixels) → image.shape[1]
    • channels → Number of color channels (3 for BGR, 1 for grayscale) → image.shape[2] (only if the image is colored)

2. image.shape[0]

  • This extracts the first element of image.shape, which is the height of the image (number of rows/pixels).
print('Height of Image: {} pixels'.format(int(image.shape[0])))
print('Width of Image: {} pixels'.format(int(image.shape[1])))
print('Depth of Image: {} colors components'.format(int(image.shape[2])))

Height of Image: 1024 pixels
Width of Image: 1024 pixels
Depth of Image: 3 colors components

1. print('Height of Image: {} pixels'.format(int(image.shape[0])))

  • image.shape[0] → Extracts the height (number of rows/pixels) of the image.
  • int(image.shape[0]) → Converts the value to an integer (although .shape always returns integers, this ensures consistency).
  • .format(int(image.shape[0])) → Inserts the height value into the {} placeholder in the string.
  • Example Output:
    Height of Image: 1024 pixels
    (if the image has 1024 rows)

2. print('Width of Image: {} pixels'.format(int(image.shape[1])))

  • image.shape[1] → Extracts the width (number of columns/pixels) of the image.
  • Works the same way as the height statement.
  • Example Output:
    Width of Image: 1024 pixels
    (if the image has 1024 columns)

3. print('Depth of Image: {} color components'.format(int(image.shape[2])))

  • image.shape[2] → Extracts the depth (number of color channels) of the image.
    • If the image is RGB/BGR, image.shape[2] = 3 (3 color channels).
    • If the image is grayscale, image.shape[2] does not exist, which would cause an IndexError.
  • Example Output (for a color image):
    Depth of Image: 3 color components
    (if the image has 3 color channels: BGR)

Wrapping Up

So, In this step by step tutorial we have learned How to,
  1. Import the OpenCV Model in Python
  2. Load Images
  3. Display Images
  4. Save Images
  5. Getting the Image Dimensions
In the end I hope you like this blog, if you are thing about learning OpenCV with Python you can start your journey with this tutorial. Thank You!
Labels:

Comments

Post a Comment