Converting list to numpy array issues

Question

Using python xy's spyder code editor.

When saving a list of pixel color values to a numpy array using np.asarray(list), it gives me an array with weird dimensions.

The dimensions of the original image being read is (864,1089) (width,height) and is a JPEG image. When converting this list to a numpy array, it returns an array with dimensions (940896, 3), when I expected a one dimensional array with dimensions (940896,). I don't understand where the 3 came from...

img = Image.open('Images\satellite6.jpg')
pix = img.load()
width = img.size[0]
height = img.size[1]
pixels = []

for j in range(height):
    for i in range(width):
        pixels.append(pix[i,j])

pixels = np.array(pixels)
print np.shape(pixels)

(940896, 3)

I am also new to posting questions, so if anything seems unclear, let me know :)

a5kin · Accepted Answer · 2014-12-14 17:02:28Z

1

Try to see what pix[0,0] contains for example:

>>> img = Image.open('./edgewalker-cat.png')
>>> pix = img.load()
>>> pix
<PixelAccess object at 0x7f1afac932f0>
>>> pix[0,0]
(0, 0, 0)

Its tuple, size of 3, so the pixels list, you are constructing one by one item, will be converted to the dimension of (width*height, 3).

UPD: On grayscaled images, you'll get a plain int value:

>>> img = Image.open('./urban-dove-gray.jpg')
>>> pix = img.load()
>>> pix[0,0]
31

edited Dec 14, 2014 at 17:02

answered Dec 14, 2014 at 16:48

a5kin

1,44518 silver badges20 bronze badges

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

2 Comments

Sean Over a year ago

pix[0,0] returns (255,255,255). What I don't understand is that I've done this with other images and it's always returned an array with one dimension. Is it because it's a bigger image?

a5kin Over a year ago

Don't think the size matters. You probably did it before with grayscaled images. Try to see the result of pix[0,0] on grayscale image - it will give you a single int.

Collectives™ on Stack Overflow

Converting list to numpy array issues

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related