将图像分配给numpy数组会使图像的颜色发生变化

Question

I load an image and add it to an empty numpy array:

plt.imshow(im)
plt.show()
im = imread('/path_to_image',mode = 'RGB')
array = np.zeros((1, 299, 299,3), dtype = np.int8)
array[0][:,:,:3] = im
print(array[0].shape)
print(im.shape)
plt.imshow(array[0])
plt.show()

I expect the two images to look the same when displayed. The image seems to change when I assign it to an array with:

array = np.zeros((1, 299, 299,3), dtype = np.int8)
array[0][:,:,:3] = im

Answer 1

im and array should have the same dtype . You can explicitly define the appropriate dtype for the array of zeros (as suggested by @Divakar) or alternatively you can use NumPy's zeros_like and newaxis as follows:

import numpy as np
import matplotlib.pyplot as plt
from skimage import io

im = io.imread('https://i.stack.imgur.com/crpfS.png')

array = np.zeros_like(im[np.newaxis, :])
array[0] = im

fig, (ax0, ax1) = plt.subplots(1, 2)
ax0.imshow(im)
ax0.set_title('im')
ax1.imshow(array[0])
ax1.set_title('array[0]')
plt.show()