In Python, how can an image stored as a NumPy array be scaled in size?

Question

I have created a NumPy array in the following way:

data = numpy.zeros((1, 15, 3), dtype = numpy.uint8)

I then filled this array with RGB pixel values, resulting in a little colour image that can be saved using a procedure such as the following:

image = Image.fromarray(data)
image.save("image.png")

How could I scale up the size of the NumPy array (without interpolation) for the purposes of creating an image that is, say, 600 x 300 pixels?

Answer 1

You can use numpy.kron as suggested in the comment or you can use the following below options

1] Using PILLOW to maintain the Aspect Ratio

• If you want to maintain the aspect ratio of the image then you can use thumbnail() method

  from PIL import Image
  
  def scale_image(input_image_path,
              output_image_path,
              width=None,
              height=None):
      original_image = Image.open(input_image_path)
      w, h = original_image.size
      print('The original image size is {wide} wide x {height} '
            'high'.format(wide=w, height=h))
  
      if width and height:
          max_size = (width, height)
      elif width:
          max_size = (width, h)
      elif height:
          max_size = (w, height)
      else:
          # No width or height specified
          raise RuntimeError('Width or height required!')
  
      original_image.thumbnail(max_size, Image.ANTIALIAS)
      original_image.save(output_image_path)
  
      scaled_image = Image.open(output_image_path)
      width, height = scaled_image.size
      print('The scaled image size is {wide} wide x {height} '
            'high'.format(wide=width, height=height))
  
  
  if __name__ == '__main__':
       scale_image(input_image_path='caterpillar.jpg',
                   output_image_path='caterpillar_scaled.jpg',
                   width=800)
  

• I used Image.ANTIALIAS flag which will apply a high quality down sampling filter which results in a better image

2] Using OpenCV

• OpenCV has cv2.resize() function

  import cv2
  image = cv2.imread("image.jpg")   # when reading the image the image original size is 150x150
  print(image.shape)
  scaled_image = cv2.resize(image, (24, 24))  # when scaling we scale original image to 24x24 
  print(scaled_image.shape)
  

• Output

  (150, 150)
  (24, 24)
  

• cv2.resize() function also has interpolation as argument by which you can specify how you want to resize the image

• INTERPOLATION METHODS:

  • INTER_NEAREST - a nearest-neighbor interpolation
  • INTER_LINEAR - a bilinear interpolation (used by default)
  • INTER_AREA - resampling using pixel area relation. It may be a preferred method for image decimation, as it gives moire’-free results. But when the image is zoomed, it is similar to the INTER_NEAREST method.
  • INTER_CUBIC - a bicubic interpolation over 4x4 pixel neighborhood
  • INTER_LANCZOS4 - a Lanczos interpolation over 8x8 pixel neighborhood

3] Using PILLOW library

• Use Image.resize()

  from PIL import Image
  sourceimage= Image.open("image.jpg")   # original image of size 150x150
  resized_image = sourceimage.resize((24, 24), resample=NEAREST)  # resized image of size 24x24
  resized_image.show()
  

4] Using SK-IMAGE library

• Use skimage.transform.resize()

  from skimage import io
  image = io.imread("image.jpg")
  print(image.shape)
  resized_image = skimage.transform.resize(image, (24, 24))
  print(resized_image.shape)
  

• Output

  (150, 150)
  (24, 24)
  

5] Use SciPy

• Use scipy.misc.imresize() function

  import numpy as np
  import scipy.misc
  
  image = scipy.misc.imread("image.jpg")
  print(image.shape)
  resized_image = scipy.misc.imresize(x, (24, 24))
  resized_image
  print(resized_image.shape)
  

• Output

  (150, 150)
  (24, 24)
  

Answer 2

In scikit-image, we have transform

from skimage import transform as tf
import matplotlib.pyplot as plt
import numpy as np
data = np.random.random((1, 15, 3))*255
data = data.astype(np.uint8)
new_data = tf.resize(data, (600, 300, 3), order=0) # order=0, Nearest-neighbor interpolation
f, (ax1, ax2, ax3) = plt.subplots(1,3, figsize=(10, 10))
ax1.imshow(data)
ax2.imshow(new_data)
ax3.imshow(tf.resize(data, (600, 300, 3), order=1))