5 Best Ways to Convert Python NumPy Array to Image

by

💡 Problem Formulation: Converting a NumPy array into an image is a common task in various fields such as data visualization, machine learning, and digital image processing. The input typically consists of a two-dimensional NumPy array for grayscale images or a three-dimensional array for color images, with each element representing a pixel value. The desired output is an image file format like PNG or JPG that graphically represents this data.

Method 1: Using the PIL/Pillow Library

PIL (Python Imaging Library) and its fork Pillow are widely used for opening, manipulating, and saving many different image file formats. The Image.fromarray() function converts a NumPy array to a PIL image object, which can then be saved to a file. This method supports various image modes (like L for grayscale and RGB for color images).

Here’s an example:

from PIL import Image
import numpy as np

# Create a NumPy array
np_array = np.random.rand(100, 100, 3) * 255
np_array = np_array.astype(np.uint8)

# Convert to an image and save
image = Image.fromarray(np_array, 'RGB')
image.save('output_image.jpg')

The snippet creates a random color image of 100×100 pixels and saves it as ‘output_image.jpg’.

This code first generates a random NumPy array with values scaled to the range of pixel values (0-255) and casts them into unsigned 8-bit integers. It then converts the array into a PIL Image object in RGB mode and saves the image to the specified location.

Method 2: Using OpenCV

OpenCV (Open Source Computer Vision Library) is a library of programming functions mainly aimed at real-time computer vision. OpenCV’s cv2.imwrite() function can directly save a NumPy array as an image, making it very convenient for image processing tasks.

Here’s an example:

import cv2
import numpy as np

# Create a NumPy array
np_array = np.random.rand(100, 100, 3) * 255
np_array = np_array.astype(np.uint8)

# Save the array as an image
cv2.imwrite('output_image.png', np_array)

The code sample generates a random image and saves it as ‘output_image.png’.

After creating and processing a NumPy array to obtain the image data, this code snippet uses OpenCV’s imwrite() function, which takes the filename where the image will be saved and the array itself as arguments, to create and save the image.

Method 3: Using Matplotlib

Matplotlib is a plotting library for the Python programming language that also has capabilities to save images. By using the function matplotlib.pyplot.imsave(), one can quickly save a NumPy array as an image file.

Here’s an example:

import matplotlib.pyplot as plt
import numpy as np

# Create a NumPy array
np_array = np.random.rand(100, 100)

# Save the array as an image
plt.imsave('output_image.png', np_array, cmap='gray')

The code creates a grayscale image and saves it as ‘output_image.png’.

This example uses Matplotlib to create a grayscale image from a 2D NumPy array. It employs the imsave() function with the ‘gray’ colormap option (cmap) which indicates the color mapping scheme for the single-channel image data.

Method 4: Using SciPy

The SciPy library provides a straightforward method to save NumPy arrays as images using the scipy.misc.imsave() function, which is however deprecated in recent versions of SciPy in favor of imageio or Pillow.

Here’s an example:

from scipy.misc import imsave
import numpy as np

# Create a NumPy array
np_array = np.random.rand(100, 100) * 255
np_array = np_array.astype(np.uint8)

# Save the array as an image
imsave('output_image.png', np_array)

This snippet saves a grayscale image as ‘output_image.png’.

This code creates a 2D NumPy array and converts it to an image using SciPy’s imsave() function. The array is converted to the appropriate data type for image representation before saving it.

Bonus One-Liner Method 5: Using Imageio

Imageio is a Python library that provides an easy interface to read and write a wide range of image data, including animated images, volumetric data, and scientific formats. The imageio.imwrite() function is a simple one-liner to save NumPy arrays as image files.

Here’s an example:

import imageio
import numpy as np

# Create a random NumPy array
np_array = np.random.rand(100, 100) * 255
np_array = np_array.astype(np.uint8)

# Save the array as an image
imageio.imwrite('output_image.png', np_array)

This code creates a grayscale image and saves it as ‘output_image.png’.

By using Imageio, the example above swiftly turns a NumPy array into an image in just one line of code. The concise syntax is perfect for quick conversions and script writing.

Summary/Discussion

  • Method 1: PIL/Pillow. Strengths: Broad image format support, intuitive API. Weaknesses: Requires an additional library beyond NumPy.
  • Method 2: OpenCV. Strengths: Fast processing for large images, good for computer vision tasks. Weaknesses: Larger library if only used for saving images.
  • Method 3: Matplotlib. Strengths: Useful if Matplotlib is already used for plotting in the project. Weaknesses: Not a primary image saving tool; more suited for visualizations.
  • Method 4: SciPy. Strengths: Part of the SciPy ecosystem, familiar syntax. Weaknesses: Deprecated in recent versions, hence requires looking at alternatives.
  • Bonus Method 5: Imageio. Strengths: Straightforward one-liner, supports a wide range of formats. Weaknesses: Additional dependency if not already in use.