How to create image from numpy float32 array?

Question

I have a numpy.ndarray array which contains values of float32. The images should be dimensions 226×226. I tried to use PIL.Image to create the image but I got an error. I read that PIL.Image.fromarray require an object and mode, and for float I need to call fromarray with 'F' as I tried.

This is what I tried to do:

from PIL import Image
img = Image.fromarray(slice56, mode='F')
#type(slice56) = <type 'numpy.ndarray'>
#slice56 = array([ 0.,  0.,  0., ...,  0.,  0.,  0.], dtype=float32)

and I get this error:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib64/python2.6/site-packages/PIL/Image.py", line 1860, in fromarray
    return frombuffer(mode, size, obj, "raw", mode, 0, 1)
  File "/usr/lib64/python2.6/site-packages/PIL/Image.py", line 1805, in frombuffer
    return apply(fromstring, (mode, size, data, decoder_name, args))
  File "/usr/lib64/python2.6/site-packages/PIL/Image.py", line 1743, in fromstring
    im = new(mode, size)
  File "/usr/lib64/python2.6/site-packages/PIL/Image.py", line 1710, in new
    return Image()._new(core.fill(mode, size, color))
TypeError: argument 2 must be sequence of length 2, not 1

Can anyone suggest an idea how to do it? Or how to solve this error?

Answer 1

I would agree with DavidG's answer being a quick solution to plot an image from a numpy array. However, if you have a very good reason for sticking with PIL.Image, the closest approach to what you've already done would be something like this:

from PIL import Image
import numpy as np

slice56 = np.random.random((226, 226))

# convert values to 0 - 255 int8 format
formatted = (slice56 * 255 / np.max(slice56)).astype('uint8')
img = Image.fromarray(formatted)
img.show()

It will then produce something like below given random numbers:

Answer 2

scipy.misc is my favorite way of doing it. It's cleaner and it is done in just one line. Look how nice it is:

import scipy.misc
img = scipy.misc.toimage(slice56, mode='L')

Please check here for the official docs.