将 2d 转换为 3d 数组后保存图像问题

Question

I have convert 2d array(arr2d) to 3d array(arr3d) and save as image using below code. arr2d is float64 type. Why save image is not color image?

arr3d[:, :, 0] = arr3d[:, :, 1] = arr3d[:, :, 2] = arr2d
arr3d=arr3d*255
im=Image.fromarray(np.maximum(np.minimum(arr3d, 255), 0).astype(np.uint8))
im.save(“sample.png”)

Answer 1

The image is not color image because for every pixel red green and blue color channels have the same value.

When using im=Image.fromarray(arr3d) :

• arr3d[:, :, 0] is the red pixels plane.
• arr3d[:, :, 1] is the green pixels plane.
• arr3d[:, :, 2] is the blue pixels plane.

When using arr3d[:, :, 0] = arr3d[:, :, 1] = arr3d[:, :, 2] = arr2d you are making sure that red=green=blue for every pixel.

The color of pixel with R=G=B is gray .

I created the following code for reproducing the issue:

import numpy as np
from PIL import Image

arr2d = np.random.rand(50, 50) # Create 50x50 2D array with random values in range [0, 1]
arr3d = np.zeros((50, 50, 3))
arr3d[:, :, 0] = arr3d[:, :, 1] = arr3d[:, :, 2] = arr2d
arr3d = arr3d*255
im = Image.fromarray(np.maximum(np.minimum(arr3d, 255), 0).astype(np.uint8))
im.save("sample.png")

Result - all pixels are gray:

Example for getting result with some colors:

arr3d[:, :, 0] = np.random.rand(50, 50);arr3d[:, :, 1] = np.random.rand(50, 50);arr3d[:, :, 2] = np.random.rand(50, 50)