[FIXED] How to make a 4D numpy array from 2 2D numpy array?

Issue

I have 2 2D numpy arrays with shape (160000,91).

160000 means all pixels of an image (200x800), 91 means # of images.

U(160000,91) and V(160000,91).

How to create a numpy array X with shape (91,200,800,2) with X(:,:,:,0) = U(91,200,800) and X(:,:,:,1) = V(91,200,800)?

Solution

Transpose to move the batch dimension to axis 0. Then reshape the two arrays and stack them along a new axis 3:

import numpy as np

U = np.random.random((160000, 91))
V = np.random.random((160000, 91))
X = np.stack((np.reshape(U.T, (-1, 200, 800)),
              np.reshape(V.T, (-1, 200, 800))),
             axis=3)
print(X.shape)  # (91, 200, 800, 2)

Answered By - asdf