Convert a PIL image to NumPy array

Key takeaways:

• Converting a PIL image to NumPy array helps in image processing and numerical analysis.

• Use pip install Pillow numpy to get Pillow for image handling and NumPy for numerical operations.

• Use Image.open('filename') from Pillow to load your image into a PIL.

• Use np.array(image) or np.asarray(image) to convert the PIL image to a NumPy array for numerical manipulation.

When working with Python, we may encounter scenarios where we have to process images and perform tasks. The Python Imaging Library (PIL) by Fredrik Lundh is used when working with images. On the other hand, NumPy helps in numerical computation and supports n-dimensional arrays.

In image processing tasks, we usually convert the images into NumPy arrays because this allows for faster and more efficient operations on large arrays and datasets. In this Answer, we’ll learn how to convert a PIL image to a NumPy array.

Install PIL and NumPy

To get started, we need to install PIL and NumPy in our system. For that, we can use the following command:

Note: Pillow is the newer version of PIL.

Once the libraries are installed successfully, we can move to the next steps.

Load the images via the Pillow library

We’ll use the following image and convert it into a NumPy array.

Now that we have installed the PIL library, we can load the image into our program. For that, we can use the following code:

Code explanation

• Line 1: We import Image module from the PIL library.

• Line 3: We open the image using the Image.open() function and pass the file name of the image as a parameter.

• Line 5: We print the image, which shows that the image is a JPEG image.

Method 1: Using np.array() to convert PIL images to NumPy array

Till now, we have installed the required libraries and loaded the image in our program. Now, we’ll convert the image to a NumPy array using the following code:

Code explanation

• Line 2: We import numpy library as np.

• Line 5: We convert the image loaded in the img variable to a NumPy array using the np.array() function.

• Line 6: We print out the shape of the NumPy array.

Method 2: Using np.asarray() to convert PIL images to NumPy array

In the code above, we can also use the np.asarray() function instead of the np.array() function. The main difference between both of the functions is that using np.array(), a copy of the original array is created, and the changes made to the original array are not reflected in the copy array. On the other hand, np.asarray() reflects all changes made to the original array in place. However, since the input is an image here and not an array, both methods will act the same way.

The NumPy array is dimensioned according to the image’s height, width, and color channels. In our case, the NumPy array’s dimension is (1600, 1200, 3), which corresponds to (the height, width, and channels) of the image.

Let’s quickly assess your understanding of converting a PIL image to a NumPy array by the following quiz:

Conclusion

Once the image has been converted to a NumPy array, we can leverage the power of the NumPy functions to perform various image processing tasks, such as filtering, blurring, and rotating an image.