Apply Canny edge filter on raster failes with scikit-image

Question

I have raster with one band and I would like to do edge detection with Canny edge filter.

This is my raster:

I have opened it with rasterio and covert it into array :

img=srs.open("raster.tif")
array=img.read()

Then i'm trying to apply the Canny filter as following (based on this: https://scikit-image.org/docs/dev/auto_examples/edges/plot_canny.html):

# Compute the Canny filter for two values of sigma
edges1 = feature.canny(array)
edges2 = feature.canny(array,sigma=3)

but it fails with this error:

ValueError: The parameter image must be a 2-dimensional array

I read that this can happpen if my image is not in grey scale, but I have shown it as grey scale:

show((img,1),cmap='Greys_r')

so I tried to checl what is my array type:

type(array)
>>>numpy.ndarray

so I have no idea why I get this error message saying the image is not 2 dimensoinal as it is 1 band with X and Y axis.

My end goal - to be able to apply Canny edge on my image and to understand why I get this error

Answer 1

Sometimes an image "looks" gray, but actually has 3 color channels with the same red, green, and blue values. To check that your image is grayscale, check array.shape rather than type(array). If it is (X, Y, 3), then you have an RGB image. To convert, you can use skimage.color.rgb2gray, and then pass the result to canny.