如何在Python中保存从灰度转换为RGB的图像？

Question

I'm new to work with Python and I'd like if there is how to save a image that I added an additional chanel.

I imported openned an image and added 2 chanels of array of zeros to try to transform to an RBG image, but the I coudn't savaet as I save usually.

from PIL import Image
import numpy as np
from array import array

i= Image.open('/content/drive/My Drive/DDICM19/imagens/mdb001.jpg')
dim = np.zeros((1024,1024))
R = np.stack((i,dim, dim), axis=2)
dim = np.zeros((1024,1024))
dim.save('/content/drive/My Drive/DDICM19/imagensP/teste.jpg')

It returned:

    AttributeError                            Traceback (most recent call last)
    <ipython-input-32-073545b24d75> in <module>()
          7 R = np.stack((i,dim, dim), axis=2)
          8 dim = np.zeros((1024,1024))
    ----> 9 dim.save('/content/drive/My Drive/DDICM19/imagensP/teste.jpg')

AttributeError: 'numpy.ndarray' object has no attribute 'save'

Somebody help.

Answer 1

First: you try to save numpy array with zeros div , not R with RGB channels.

---

But R is numpy array and you have to convert it back to PIL image

Image.fromarray(R, 'RGB').save('output.jpg')

To convert grayscale to RGB better repeat the same values for R, G, B instead of adding zeros

R = np.stack((i, i, i), axis=2)

With zeros it gives me something strange. It has to uses int8 or unit8 data type to correctly convert it to RGB

dim = np.zeros((i.size[1], i.size[0]), 'uint8')

I also use size from image to create array with zeros but you have to remeber that image uses (x,y) and array use (y, x) which means (row, column)

---

Example

from PIL import Image
import numpy as np

i = Image.open('image.jpg')
#i = i.convert('L') # convert RGB to grayscale to have only one channel for tests
print(i.size)  # (x, y)

dim = np.zeros((i.size[1], i.size[0]), 'int8') # array uses different order (y, x)
print(dim.shape)

R = np.stack((i, dim, dim), axis=2)
#R = np.stack((i, i, i), axis=2) # convert grayscale to RGB
print(R.shape)
#print(R[0,2]) # different values if not used `int8` #(y,x)

img = Image.fromarray(R, 'RGB')
img.save('output.jpg')
#print(img.getpixel((2,0))) # different values if not used `int8` #(x,y)

---

EDIT: You can also convert array of zeros to grayscale image which can be used as channel in image

 img_zero = Image.fromarray(dim, 'L')

 img = Image.merge('RGB', (i, img_zero, img_zero))
 img.save('output.jpg')

and then image looks interesting.