How to resize image without adding noise with python?

Question

I have an image and I want to resize and expand this image without spoiling it.

The image:

The image size:

w=41
h=43

The image size I want it resized to:

w=2000x
h=2100

I tried all the known methods:

with PIL.Image:

from PIL import Image
path = "image.png"
w, h = 2000, 2100
img = Image.open(path)
img = img.resize((w, h), Image.ANTIALIAS)
img.save("re_image.png", "PNG")

with cv2:

import cv2
path = "image.png"
w, h = 2000, 2100
img = cv2.imread(path)
img = cv2.resize(img, (w, h), interpolation = cv2.INTER_AREA)
cv2.imwrite("re_image.png", img)

The result is:

It depends on what you want the final image to look like. If you want it to keep it's pixelation the cv2.INTER_NEAREST is probably your best bet. All methods of interpolating between existing pixels will create some artefact. Especially at the scales you are attempting. — BHawk
– BHawk, Commented Jan 15, 2020 at 12:39

sshashank124 · Accepted Answer · 2020-01-15 12:44:54Z

6

Don't use the ANTIALIAS filter. Since you want to preserve the sharpness of the edges, simply use the NEAREST filter. This filter is the default if no argument is passed to the resample parameter of Image.resize

img = img.resize((w, h))

This gives the expected result

Expected Result

edited Jan 15, 2020 at 12:44

answered Jan 15, 2020 at 12:39

sshashank124

32.3k10 gold badges72 silver badges76 bronze badges

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

Comments

87VN0 · Accepted Answer · 2020-01-15 12:40:20Z

1

For your specific case which is a binary image, I think that you want to keep the edges sharp. Then I would suggest to use the flag cv2.INTER_NEAREST in your openCV implementation. You can check the effect of the different interpolation types, given in the documentation: https://docs.opencv.org/4.1.1/da/d54/group__imgproc__transform.html

I hope this helps!

answered Jan 15, 2020 at 12:40

87VN0

7953 silver badges10 bronze badges

1 Comment

hsarret Over a year ago

If you use Pillow, it looks like NEAREST is not default so you have you use img = img.resize((w, h), PIL.Image.NEAREST)

Gino Mempin · Accepted Answer · 2020-01-16 04:16:46Z

-1

I created a simple function. Please give me your opinion.

The function is:

 import cv2, numpy as np

 def resize(path, sizes = (500, 800)):
         nw, nh = sizes
         g_img = cv2.imread(path)

         gw, gh = g_img.shape[0], g_img.shape[1]
         nw, nh = gw+510, gh+510
         pw, ph = int(nw/gw), int(nh/gh)
         if (pw == 0):pw = 1
         if (ph == 0):ph = 1
         nw, nh = pw*gw, ph*gh

         cvimg = np.zeros((nh,nw,3), np.uint8)

         for gy in range(gh):
             y, h = gy*ph, (gy*ph)+ph
             for gx in range(gw):
                 x, w = gx*pw, (gx*pw)+pw
                 cvimg[y:h, x:w] = g_img[gy:gy+1, gx:gx+1]

         nsize = [nw, nh]
         return cvimg, nsize

The result is:

edited Jan 16, 2020 at 4:16

Gino Mempin

30.5k31 gold badges125 silver badges174 bronze badges

answered Jan 15, 2020 at 15:52

abdo Information

732 silver badges9 bronze badges

Collectives™ on Stack Overflow

How to resize image without adding noise with python?

3 Answers 3

Comments

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related