CODEWITHSUNDEEP
pythonarraysnumpyscikit-imagecanny-operator
Simon Steinberger
Simon Steinberger

Reputation: 6815

Read image file in Python and compute Canny edge filters

I'd like to read an image file with Python and apply skimage's Canny edge filter. But I don't know the correct array format for the feature computation. That's what I have:

from PIL import Image
from skimage import feature

PATH = '/foo/bar.jpg'

import numpy
img = numpy.asarray(Image.open(PATH).convert('LA'))

# apply Canny Edge filters
edges1 = feature.canny(img)
edges2 = feature.canny(img, sigma=3)

The feature call raises this error: "The parameter image must be a 2-dimensional array". How do I convert the numpy array to the necessary form?

Upvotes: 1

Views: 3775

Answers (1)

kmario23
kmario23

Reputation: 61325

From the description of your question, it seems that you're dealing with RGB images (i.e. color images). For such images, we have to first convert it to a grayscale image and then only we can pass them to Canny Edge Detector because the parameter image has to be a 2D array.

image : 2D array
Greyscale input image to detect edges on; can be of any dtype.

Here is an example:

# load color image
In [12]: img_rgb = 'model.jpg'
In [13]: img_arr = np.array(Image.open(img_rgb), dtype=np.uint8)

In [14]: img_arr.shape
Out[14]: (1005, 740, 3)

# convert to grayscale image
In [15]: from skimage.color import rgb2gray
In [16]: img_gray = rgb2gray(img_arr)

In [17]: img_gray.shape
Out[17]: (1005, 740)

In [18]: edges1 = feature.canny(img_gray)
    ...: edges2 = feature.canny(img_gray, sigma=3)

In [19]: edges1.shape
Out[19]: (1005, 740)

In [20]: edges2.shape
Out[20]: (1005, 740)

# display    
In [21]: plt.imshow(edges1)

And I get the result as in the below image:

model image

Upvotes: 5

Related Questions