3

I have a greyscale image represented as 2D numpy array and want to make it a 3D numpy array representing a color rgb image (which is obviously still grey).

img.shape       // (100, 100)
img[10, 10]     // e.g. 42

// do something

img.shape       // (100, 100, 3)
img[10, 10]     // e.g. [42, 42, 42]

I found this question which asked the opposite: numpy 3D-image array to 2D

QuantumX
  • 33
  • 5

3 Answers3

3

You can use Numpy's dstack() to stack three grey images depthwise:

RGB = np.dstack((grey, grey, grey))
Mark Setchell
  • 191,897
  • 31
  • 273
  • 432
0

You could use an opencv function for this: image_color = cv2.cvtColor(image_gray, colorcode).

See the available colorcodes in documentation https://docs.opencv.org/3.4/d8/d01/group__imgproc__color__conversions.html

The right one for you could be cv2.COLOR_GRAY2RGB.

import cv2
image_color = cv2.cvtColor(image_gray, cv2.COLOR_GRAY2RGB)

That would give you the grayscale image as 3 channel rgb array.

Using numpy you could look at https://stackoverflow.com/a/40119878/14997346

chillking
  • 311
  • 1
  • 9
0

Does the following code help?

import numpy as np
aa = np.zeros((4,4),dtype='int64')
aa[1,1] = 42
aa = aa.reshape((4,4,1))
bb = np.broadcast_to(aa,(4,4,3))
print(bb[1,1,:])
NNN
  • 697
  • 1
  • 5
  • 15