Not so canny canny method?

[JobJob]

img = rand(RGB{N0f8}, 100, 100)
canny(img, (Percentile(80), Percentile(20)))

Gives a

ERROR: MethodError: no method matching canny(::Array{Gray{Normed{UInt8,8}},2})
Stacktrace:
 [1] canny(::Array{RGB{Normed{UInt8,8}},2}, ::Tuple{Percentile{Int64},Percentile{Int64}}) at Images/7FSUM/src/edge.jl:410

Seems this code just needs to forward the threshold argument:

Images.jl/src/edge.jl

Lines 410 to 411 in 3951d87

	canny(img::AbstractMatrix, threshold::Tuple{N,N}, args...) where {N<:Union{NumberLike,Percentile{NumberLike}}} =
	canny(convert(Array{Gray}, img), args...)

i.e. canny(convert(Array{Gray}, img), threshold, args...) though a warning suggests that maybe it should be canny(Gray.(img), threshold, args...)

[zygmuntszpak]

Thanks for reporting this. You are right, line 411 should be forwarding the threshold argument.

[johnnychen94]

@zygmuntszpak Given that ImageEdgeDetection already implements Canny, what's the current status of this issue? My guess is that we only need to properly deprecate the old canny in src/edge.jl in favor of ImageEdgeDetection, and then this issue gets solved?

[ashwanirathee]

thinning algorithms have also been implemented in ImageEdgeDetection.jl and they also need to be properly deprecated here in Images.jl I think

[Devanik21]

To fix the error you're encountering with the canny function, you need to convert the image to grayscale before applying the Canny edge detection algorithm. Here are steps-

Convert the RGB image to grayscale.
Apply the Canny edge detection algorithm to the grayscale image.

[Devanik21]

#try this code, it can fix the error
using Images
using ImageFiltering
img = rand(RGB{N0f8}, 100, 100)
gray_img = Gray.(img)

canny_img = canny(gray_img, (Percentile(80), Percentile(20)))