Reputation: 145
There are two situations when I open tif image by PIL. First one is a normal one that can open an image as RGB and display by plt well. However, the second one as shown in the codes below:
from PIL import Image
import numpy as np
import matplotlib.pyplot as plt
img = Image.open('test1.tif')
a = np.array(img)
a.shape
# output: (2040,2040)
can only open as a shape of (2040,2040), the previous one with a shape of (2040,2040,3). When I convert the (2040,2040) one into RGB mode using img = img.convert('RGB')
and showing it via plt only saw a blank square.
Please, somebody, tell me what is going wrong? Thanks in advance!
And numpy array:
array([[7256, 6926, 7270, ..., 7368, 7457, 7555],
[7174, 6945, 6878, ..., 7401, 7353, 7895],
[7288, 7099, 7140, ..., 7359, 7524, 7587],
...,
[6769, 6695, 6698, ..., 6599, 6788, 6898],
[6780, 6683, 6857, ..., 6723, 6761, 6861],
[6788, 6626, 6761, ..., 6738, 6751, 7054]], dtype=uint16)
After convert to RGB mode:
numpy array:
array([[[255, 255, 255],
[255, 255, 255],
[255, 255, 255],
...,
[255, 255, 255],
[255, 255, 255],
[255, 255, 255]]], dtype=uint8)
Upvotes: 3
Views: 6214
Reputation: 3491
The issue you are having is because the image is a 16-bit image (uint16) which is unsupported by the PIL image library. You can, as you suggested, simply convert it to a different image type. However, you can also just use imageio
The following code should work:
import imageio
import matplotlib.pyplot as plt
%matplotlib inline
img = imageio.imread('test1.tif')
plt.imshow(img)
See here for a discussion of related topics: https://github.com/imageio/imageio/issues/204
BTW, here's a list of PIL image formats and how many channels they have (Adapted from https://helpful.knobs-dials.com/index.php/Python_usage_notes_-_PIL):
PIL pixel formats:
RGB 24bits per pixel, 8-bit-per-channel RGB), 3 channels
RGBA (8-bit-per-channel RGBA), 4 channels
RGBa (8-bit-per-channel RGBA, remultiplied alpha), 4 channels
1 - 1bpp, often for masks, 1 channel
L - 8bpp, grayscale, 1 channel
P - 8bpp, paletted, 1 channel
I - 32-bit integers, grayscale, 1 channel
F - 32-bit floats, grayscale, 1 channel
CMYK - 8 bits per channel, 4 channels
YCbCr - 8 bits per channel, 3 channels
Upvotes: 5