1

I'm trying to compute grey level co-occurrence matrices from images for feature extraction. I'm using greycomatrix for the task but there seems to be something I don't understand about the process since I'm getting the following error:

ValueError: buffer source array is read-only

(The full trace can be found below)

So here's what I've done:

Converting the (PIL) image to grayscale with 8 quantization levels:

greyImg = img.convert('L', colors=8)

And then compute the glcm matrices:

glcm = greycomatrix(greyImg, distances=[1], angles=[0, np.pi/4, np.pi/2], 
                    symmetric=True, normed=True)

This results in a rather cryptic error:

glcm = greycomatrix(img, distances=[1], angles=[0, np.pi/4, np.pi/2], levels=256, symmetric=True, normed=True)

_glcm_loop(image, distances, angles, levels, P)

File "skimage/feature/_texture.pyx", line 18, in skimage.feature._texture._glcm_loop

File "stringsource", line 654, in View.MemoryView.memoryview_cwrapper

File "stringsource", line 349, in View.MemoryView.memoryview._cinit__ ValueError: buffer source array is read-only

I've been trying to tingle with the paremeters but I can't seem to figure out, why this happens. What would be the correct way to compute the glcm-matrix?

Update

The problem was in the grayscale conversion. The following changes were required: 

import numpy as np

greyImg = np.array(img.convert('L', colors=8))
Iizuki
  • 354
  • 1
  • 6
  • 12

1 Answers1

0

The function greycomatrix expects a NumPy ndarray rather than a PIL Image object. You need to convert greyImg like this:

import numpy as np

greyImg = np.asarray(img.convert('L', colors=8))
Tonechas
  • 13,398
  • 16
  • 46
  • 80
  • 1
    And you probably want to read the image using `skimage.io`. Doing conversion from PIL by hand is brittle. – Stefan van der Walt Apr 23 '19 at 18:27
  • Yes, that was my problem. Thank you. However, `np.asarray()` produced the exact same error. `np.array` did the trick. – Iizuki Apr 28 '19 at 12:15
  • Either of them worked fine in my computer. Take a look at [this post](https://stackoverflow.com/questions/14415741/numpy-array-vs-asarray) – Tonechas Apr 28 '19 at 18:44