B&W image to binary array

Question

I want to convert my b&w image(.png) to binary array(black is 1 white is 0). I have written some code, but it's not working. Error says: argument 2 to map() must support iteration. Here is my code:

from PIL import Image
from resizeimage import resizeimage
import sys

def threshold(col):
    s = sum(col) 
    return int(s > 255 * 3 // 2)

img = Image.open("filename.png")

ratio = float((img.size[1]) / (img.size[0]))

img = resizeimage.resize_cover(img, [100, int(ratio * 100)])

pixels = img.getdata()
binary = list(map(threshold, img))

array2d = [binary[i * img.size[0] : (i+1) * img.size[0]] for i in range(img.size[1])]
print('\n'.join(''.join(map(str, line)) for line in array2d))

Here is the image:

example image

no only b&w, image is just I drew a black circle with paint, with white backgound. — Hasmik Khachatryan, May 23 '18 at 17:57
Replace `binary = list(map(threshold, img))` with `binary = list(map(threshold, pixels))` — 3Dave, May 23 '18 at 18:21
What about converting to a NumPy array with `np_img = np.array(img)`, then setting non-zero values to 1 `np_img[np_img > 0] = 1` — Fiver, May 23 '18 at 18:50
With that, when I print np_img, I see just 1. And I tried to convert ndarray to list using `np_img.tolist()` , but it's not converting. — Hasmik Khachatryan, May 23 '18 at 19:31
I did, you can check it out, its just black circle, that I drew with paint — Hasmik Khachatryan, May 23 '18 at 19:41

Fiver · Accepted Answer · 2018-05-23T22:45:21.027

2

You need to convert your image to grayscale first, since PIL opens it as RGB. Then, invert the 0 & 255 values. Then, you can convert the non-zero values to 1. Here's one way:

from PIL import Image
import numpy as np

img = Image.open('bw_circle.png').convert('L')

np_img = np.array(img)
np_img = ~np_img  # invert B&W
np_img[np_img > 0] = 1

And an alternative way using PIL for the inversion:

from PIL import Image, ImageOps
import numpy as np

img = Image.open('bw_circle.png').convert('L')
img_inverted = ImageOps.invert(img)

np_img = np.array(img_inverted)
np_img[np_img > 0] = 1

edited May 23 '18 at 22:45

answered May 23 '18 at 19:55

Fiver

9,909
9
43
63

I printed np_img [[1 1 1 ..., 1 1 1] [1 1 1 ..., 1 1 1] [1 1 1 ..., 1 1 1] ..., [1 1 1 ..., 1 1 1] [1 1 1 ..., 1 1 1] [1 1 1 ..., 1 1 1]] – Hasmik Khachatryan May 23 '18 at 20:16
instance of 1 i need to print 0 . same for 0 to 1 .How will do that . – Boxer Robert Jan 05 '19 at 01:09

score 0 · Answer 2 · answered Jul 26 '23 at 23:33

0

You could also do this:

from PIL import Image
import numpy as np

image = Image.open('bw_circle.png').convert('1')
binary = bytes(np.packbits(~np.array(image)))

answered Jul 26 '23 at 23:33

panda-byte

463
3
10

B&W image to binary array

2 Answers2