How to save and show 2d numpy array as an RGB pseudo colored image?

You could use PIL.Image to do that, but first transform your array.

You could say for example, that:

1. 1.0 Should be red, represented as (255,0,0)
2. 2.0 Should be green -> (0,255,0)
3. 3.0 Should be blue -> (0,0,255)
4. 4.0 Should be black -> (0,0,0)
5. 5.0 Should be white -> (255,255,255)

You could of course change those values to whatever colors you choose, but this is just for demonstration. That being said, your 2-d array also needs to be "flattened" to 1-d for PIL.Image to accept it as data.

from PIL import Image
import numpy as np
your_2d_array = np.something() # Replace this line, obviously
img_array = []
for x in your_2d_array.reshape(2000*2000):
    if x == 1.0:
        img_array.append((255,0,0)) # RED
    elif x == 2.0:
        img_array.append((0,255,0)) # GREEN
    elif x == 3.0:
        img_array.append((0,0,255)) # BLUE
    elif x == 4.0:
        img_array.append((0,0,0)) # BLACK
    elif x == 5.0:
        img_array.append((255,255,255)) # WHITE

img = Image.new('RGB',(2000,2000))
img.putdata(img_array)
img.save('somefile.png')

While this should work, I think there are more efficient ways to do this that I do not know, So I will be glad if someone edits this answer with better examples. But if it's a small app and maximum efficiency doesn't bother you, here it is.

matplotlib is useful for tasks like this, though there are other ways. Here's an example:

import numpy as np
import matplotlib.image

src = np.zeros((200,200))
print src.shape

rgb = np.zeros((200,200,3))
print rgb.shape

src[10,10] = 1
src[20,20] = 2
src[30,30] = 3

for i in range(src.shape[0]):
  for j in range(src.shape[1]):
    rgb[i,j,0] = 255 if src[i,j]==1 else 0 # R
    rgb[i,j,1] = 255 if src[i,j]==2 else 0 # G
    rgb[i,j,2] = 255 if src[i,j]==3 else 0 # B

matplotlib.image.imsave('test.png', rgb.astype(np.uint8))

The trick is to convert it to an RGB array of shape (x, y, 3). You can use any formula you want to gen the per-pixel RGB values.

Also, note converting it to an uint8 array.