Image, CV2 read the conversion and size resize change of numpy array of pictures

Several image size modifications and parameter summaries

（from torchvision import transforms as T）

Display different size formats

• Image The type and T Conduct resize The picture of size attribute The order of display parameters is W、H
• cv2 Display of Of shape The order of display parameters is H、W、C
• T.ToTensor()(img), Convert the picture to tensor type , Display is size() Method , for example ,

import numpy as np
from PIL import Image
from torchvision import transforms as T
# import matplotlib.image as Img

img = Image.open('4.jpg')
img_tran = T.Resize((512,256))(img)
print(img_tran.size)   #(256, 512)
print(np.array(img_tran).shape)  # (512, 256, 3)
img_to = T.ToTensor()(img_tran)
img_to_array = np.array(img_to)
print(img_to.size()) #torch.Size([3, 512, 256])
print(img_to_array.shape) # (3, 512, 256)

• It can be seen that size() Method shows that Channel comes first 了 .

Modify the function parameter order of image size

-T.Resize((H,W))(img)

• img.resize((W,H))、cv2.resize(img_read,(W,H))

Image Picture data and numpy Data conversion

Image Convert the picture to numpy data

np.array(img)

from PIL import Image
import numpy as np
img = Image.open('4.jpg')
img_array = np.array(img)

numpy Data to Image picture

Image.fromarray(img_arr.astype(‘uint8’))
Image.fromarray(np.uint8(img))

from PIL import Image
import numpy as np
img = Image.open('4.jpg')
img_array = np.array(img)
img_image = Image.fromarray(img_arr.astype('uint8'))

Image Dimension display and numpy Of shape Show problems

Image.size attribute It is shown that ** wide 、 high **
img_array.shape attribute , It is shown that ** high 、 wide 、 passageway **

from PIL import Image
from torchvision import transforms as T
import numpy as np
img = Image.open('4.jpg')
img_array = np.array(img)
print(img.size)   # (720, 1160)
print(img_arr.shape)  # (1160, 720, 3)

plt.figure()

plt.subplot(1,3,1)
plt.imshow(img)

plt.subplot(1,3,2)
plt.imshow(img_array)
plt.show()

img.resize（w, h）,T.Resize((h,w))(img),cv2.resize(img,(w,h))

function ： Modify the width and height of the read image size

import cv2
import matplotlib.image as Img
from torchvision import transforms as T

img = Image.open('4.jpg')
img_resize = img.resize((512,256))
img_tran = T.Resize((512,256))(img)
img_array = np.array(img)

img_read = cv2.imread('4.jpg')
img_cv2 = cv2.resize(img_read,(512,256))

print(img_resize.size)  #(512, 256)
print(img_array.shape)  #(1160, 720, 3)
print(img_tran.size)  # (256, 512)
print(img_cv2.shape)  # (256, 512, 3)

plt.figure()

plt.subplot(1,3,1)
plt.title('img_resize')
plt.imshow(img_resize)

plt.subplot(1,3,2)
plt.title('img_tran')
plt.imshow(img_tran)

plt.subplot(1,3,3)
plt.title('img_cv2')
plt.imshow(img_cv2)

plt.show()