How would I scale a 2-dimensional array in python?

Question

I'm not sure how to go about scaling a 2-dimensional array. Given the array below, whose dimensions are 8x10, say I needed to scale it to 5x6 -- I've looked for concrete examples on wikipedia, but without much grounding in matrix math I'm a bit lost. If someone could point me in the right direction I'd really appreciate it!

[
 [0, 0, 1, 1, 1, 1, 0, 0],
 [0, 1, 1, 1, 1, 1, 1, 0],
 [0, 1, 0, 0, 0, 1, 1, 1],
 [0, 0, 0, 0, 0, 0, 1, 1],
 [0, 0, 1, 1, 1, 1, 1, 1],
 [1, 1, 1, 1, 1, 1, 1, 1],
 [1, 1, 0, 0, 0, 0, 1, 1],
 [1, 1, 0, 0, 0, 1, 1, 1],
 [1, 1, 1, 1, 1, 1, 1, 1],
 [0, 1, 1, 1, 1, 0, 1, 1]
]

When you say scale, what exactly do you mean? Treat it as a picture? — configurator
– configurator, Commented Dec 15, 2010 at 6:39
Scaling has a specific meaning with regards to matrices (namely, it's a particular affine transformation), which does not change the size of the matrix. Thus I guess you're referring to something else here, but it's not clear what that is. — Adrian Petrescu
– Adrian Petrescu, Commented Dec 15, 2010 at 6:46
Yah, treat it like a picture. I think this is where my problems finding a solution have stemmed because matrix transformation seems to be a bit different. I'd like to "preserve" in some sense the data contained within while scaling down or up arbitrarily — coleifer
– coleifer, Commented Dec 15, 2010 at 13:53

Justin Peel · Accepted Answer · 2010-12-15 07:13:33Z

10

Since your array looks like it is a binary image of a lower case 'a' letter, I'm guessing that you mean scaling in the image sense.

To do that, I would recommend using the imresize function in scipy.misc (which is taken from PIL, I believe). Here's an example:

import numpy as np
from scipy.misc import imresize

img = np.array([
 [0, 0, 1, 1, 1, 1, 0, 0],
 [0, 1, 1, 1, 1, 1, 1, 0],
 [0, 1, 0, 0, 0, 1, 1, 1],
 [0, 0, 0, 0, 0, 0, 1, 1],
 [0, 0, 1, 1, 1, 1, 1, 1],
 [1, 1, 1, 1, 1, 1, 1, 1],
 [1, 1, 0, 0, 0, 0, 1, 1],
 [1, 1, 0, 0, 0, 1, 1, 1],
 [1, 1, 1, 1, 1, 1, 1, 1],
 [0, 1, 1, 1, 1, 0, 1, 1]
])
newimg = imresize(img, (6,5))

and newimg is then

array([[  0,   0, 255, 255,   0],
       [  0, 255, 255, 255, 255],
       [  0,   0,   0,   0, 255],
       [255, 255, 255, 255, 255],
       [255, 255,   0,   0, 255],
       [255, 255, 255, 255, 255]], dtype=uint8)

which isn't perfect, but you can change the 255's to 1's easily enough. Also, if you get the development version of Scipy, currently version 9, then you have some other parameters(scroll down to imresize - no anchor) that you can input to imresize such as the interpolation method and PIL mode.

answered Dec 15, 2010 at 7:13

Justin Peel

47.1k6 gold badges62 silver badges81 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

martineau Over a year ago

Why use scipy when just using PIL directly makes more sense?

coleifer Over a year ago

Wow, I totally figured it would a hack to use PIL directly, but maybe that's the best way to go

Justin Peel Over a year ago

@martineau, maybe because you want to use the result back in scipy? That was my thought anyway.

Community · Accepted Answer · 2017-05-23 10:31:04Z

2

When in doubt, use a library!

PIL has functions for resizing images, assuming that's the kind of scaling you're after. To convert your list of lists to a PIL image, first convert it to a numpy array, then to PIL format. After you're done, you can reverse the process to get a list of lists again (if you really want).

edited May 23, 2017 at 10:31

CommunityBot

11 silver badge

answered Dec 15, 2010 at 7:07

perimosocordiae

17.9k14 gold badges64 silver badges76 bronze badges

Collectives™ on Stack Overflow

How would I scale a 2-dimensional array in python?

2 Answers 2

3 Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related