How to blur an image in Python pillow
Pillow is a fork for the Python Imaging Library, PIL, that provides various image-processing features to manipulate images and apply different filters. It is a powerful library that is easy to integrate with other Python libraries and frameworks. By using the pillow library, we can perform various operations on an image as follows:
Filter effects like blurring
Image enhancement, like adjusting brightness, contrast, and sharpness
Image overlay, like adding text on an image
In this Answer, we shall apply different types of blur filter effects on an image and observe how the image changes.
Required imports
We will use the PIL library to manipulate the image and then use matplotlib to display the results once the code is executed successfully.
import matplotlib.pyplot as pltfrom PIL import Image, ImageFilterimport requestsfrom io import BytesIO
matplotlib: To create visualizations and plots in the Python library.PIL: Computer vision library to apply operations on images.Image: To manipulate the images.ImageFilter: To access different filters, including blur.
requests: To send the requests over the HTTP server to the website.BytesIO: To handle the binary data as an in-memory system.
Apply simple blur
A simple blur is an image-blurring technique that replaces the image pixel values with the mean value of the neighboring pixel. This reduces the image's noise and decreases the edges' sharpness, resulting in a comparatively smoother appearance.
Expected output
When a simple blur is applied to the original picture, its overall resolution decreases as all the pixels are changed to the average pixel value, and the appearance is more smooth.
Example code
In this example, we fetch an online image through its link address and apply a simple blur effect by using ImageFilter.BLUR.
import matplotlib.pyplot as plt
from PIL import Image, ImageFilter
import requests
from io import BytesIO
image_url = "https://images.pexels.com/photos/699466/pexels-photo-699466.jpeg?auto=compress&cs=tinysrgb&w=300"
#fetch image
response = requests.get(image_url)
image_data = BytesIO(response.content)
original_image = Image.open(image_data)
#Apply simple blur
blur_image = original_image.filter(ImageFilter.BLUR)
plt.figure(figsize=(10, 5))
plt.subplots_adjust(wspace=0.4, hspace=0.1)
plt.subplot(1, 2, 1)
plt.imshow(original_image)
plt.axis('off')
plt.title('Original Image')
plt.subplot(1, 2, 2)
plt.imshow(blur_image)
plt.axis('off')
plt.title('Blurred Image')
plt.show()
Code explanation
Lines 1–4: Import the required libraries and modules.
Line 6: Store the link of the image that is to be used on the
image_url. We can also use a local image file and give its exact name.Lines 9–11: Use
requestto fetch the image from the server and open it using theImage.open()method and pass the converted image to it.Line 14: Use the
filtermethod ofImageclass to apply theBLUR()effect.Lines 16–17: Specify the figure size that is to appear in the plot and assign the grid size.
Lines 20–23: Display the image using
imshow()and pass the original image to it, turn off the x-axis labels, and specify the title in thetitle().Lines 25–28: Display the image using
imshow()and pass the blurred image to it, turn off the x-axis labels, and specify the title in thetitle().Line 30: Show the resultant plot.
Apply box blur
Now let's blur the image more by applying a box blur. A box blur is an image-blurring technique that applies the concept of replacing the image pixel values with the mean value of the neighboring pixel. However, it uses a kernel to apply the mean value concept with the same value for all the pixels in a specified box-shaped area. This provides an even more smooth appearance.
Expected output
When a box blur is applied to the original picture, its overall smoothness improves, and the details of the image are not clearly visible.
Example code
In this example, we fetch an online image through its link address and apply a box blur effect by using ImageFilter.BoxBlur(6)) where 6 is the radius value.
import matplotlib.pyplot as plt
from PIL import Image, ImageFilter
import requests
from io import BytesIO
image_url = "https://images.pexels.com/photos/699466/pexels-photo-699466.jpeg?auto=compress&cs=tinysrgb&w=300"
#fetch image
response = requests.get(image_url)
image_data = BytesIO(response.content)
original_image = Image.open(image_data)
# Apply box blur
blur_image = original_image.filter(ImageFilter.BoxBlur(6))
plt.figure(figsize=(10, 5))
plt.subplots_adjust(wspace=0.4, hspace=0.1)
plt.subplot(1, 2, 1)
plt.imshow(original_image)
plt.axis('off')
plt.title('Original Image')
plt.subplot(1, 2, 2)
plt.imshow(blur_image)
plt.axis('off')
plt.title('Box blurred Image')
plt.show()Code explanation
Lines 1–4: Import the required libraries and modules.
Line 6: Store the link of the image that is to be used on the
image_url. We can also use a local image file and give its exact name.Lines 9–11: Use
requestto fetch the image from the server and open it using theImage.open()method and pass the converted image to it.Line 14: Use the
filtermethod ofImageclass to apply theBoxBlur()and pass the radius value as a parameter.Lines 16–17: Specify the figure size that is to appear in the plot and assign the grid size.
Lines 20–23: Display the image using
imshow()and pass the original image to it, turn off the x-axis labels, and specify the title in thetitle().Lines 25–28: Display the image using
imshow()and pass the blurred image to it, turn off the x-axis labels, and specify the title in thetitle().Line 30: Show the resultant plot.
Apply Gaussian blur
Now let's achieve a smoother appearance and reduce the noise by applying Gaussian blur. A Gaussian blur is an image-blurring technique that applies a Gaussian distribution function on the pixels in the specified kernel area. This approach gives more weight to the pixels in the center, enabling a blur effect without losing the image definition.
Expected output
When a Gaussian blur is applied to the original picture, the details of the image are not visible, and a more smooth image is because the noise is reduced, but notice that the edges are still preserved.
Example code
In this example, we fetch an online image through its link address and apply a Gaussian blur effect by using ImageFilter.GaussianBlur(6)) where 6 is the radius value.
import matplotlib.pyplot as plt
from PIL import Image, ImageFilter
import requests
from io import BytesIO
image_url = "https://images.pexels.com/photos/699466/pexels-photo-699466.jpeg?auto=compress&cs=tinysrgb&w=300"
#fetch image
response = requests.get(image_url)
image_data = BytesIO(response.content)
original_image = Image.open(image_data)
#Apply guassian blur
blur_image = original_image.filter(ImageFilter.GaussianBlur(6))
plt.figure(figsize=(10, 5))
plt.subplots_adjust(wspace=0.4, hspace=0.1)
plt.subplot(1, 2, 1)
plt.imshow(original_image)
plt.axis('off')
plt.title('Original Image')
plt.subplot(1, 2, 2)
plt.imshow(blur_image)
plt.axis('off')
plt.title('Blurred Image')
plt.show()
Code explanation
Lines 1–4: Import the required libraries and modules.
Line 6: Store the link of the image that is to be used on the
image_url. We can also use a local image file and give its exact name.Lines 9–11: Use
requestto fetch the image from the server and open it using theImage.open()method and pass the converted image to it.Line 14: Use the
filtermethod ofImageclass to apply theGaussianBlur()and pass the radius value as a parameter.Lines 16–17: Specify the figure size that is to appear in the plot and assign the grid size.
Lines 20–23: Display the image using
imshow()and pass the original image to it, turn off the x-axis labels, and specify the title in thetitle().Lines 25–28: Display the image using
imshow()and pass the blurred image to it, turn off the x-axis labels, and specify the title in thetitle().Line 30: Show the resultant plot.
Real-life application
There are a lot of real-life scenarios where there is a need to blur an image. Let's take a look at few of the scenarios where blurring an image is useful.
Summary
Image blur is one of the commonly used filter effects when it comes to images and can be helpful to cater to a lot of real-life needs. We can apply a simple, box, and Gaussian blur and achieve different kinds of blur effects depending on the requirement.
Free Resources
