如何像2D一样转置3D矩阵？

Question

I want to transpose 3D matrix but without touching most inner list. The 3D matrix is numpy array of an image and the goal of transposition is to rotate image but not to change pixel values(rgb).

image = np.array(
    [
        [[15, 15, 25], [50, 50, 50], [90, 90, 90], [255, 255, 255], [127, 127, 127]],

        [[121, 121, 121], [75, 10, 75], [187, 187, 187], [0, 0, 0], [210, 50, 100]],

        [[31, 31, 31], [63, 63, 63], [87, 87, 87], [63, 63, 63], [63, 63, 63]],

        [[95, 95, 95], [95, 95, 95], [115, 115, 115], [87, 87, 87], [95, 95, 95]],

        [[95, 95, 95], [63, 63, 63], [63, 63, 63], [87, 87, 87], [100, 250, 250]]
    ], dtype=np.uint8)

The result should be something like this.

result = np.array(
    [
        [[15, 15, 25], [121, 121, 121], [31, 31, 31], [95, 95, 95], [95, 95, 95]],

        [[50, 50, 50], [75, 10, 75], [63, 63, 63], [95, 95, 95], [63, 63, 63]],

        [[90, 90, 90], [187, 187, 187], [87, 87, 87], [115, 115, 115], [63, 63, 63]],

        [[255, 255, 255], [0, 0, 0], [63, 63, 63], [87, 87, 87], [87, 87, 87]],

        [[127, 127, 127], [210, 50, 100], [63, 63, 63], [95, 95, 95], [100, 250, 250]]
    ], dtype=np.uint8)

Answer 1

看看np.swapaxis

np.swapaxes(image, 0, 1)

Answer 2

I think the answer to this can be found here

basically use this

numpy.swapaxes(image,0,1)