Where/If to add these functions?

[aquatiko]

I noticed in JuliaImages doc that the function find_boundaries is not listed for Julia. I would like to implement it(If not already implemented in Julia).

Also, for it's purpose I would like to implement something similar to generate_binary_structures from scipy https://github.com/scipy/scipy/blob/f2ec91c4908f9d67b5445fbfacce7f47518b35d1/scipy/ndimage/morphology.py#L122. And, numpy.indices from https://github.com/numpy/numpy/blob/ccfbcc1cd9a4035a467f2e982a565ab27de25b6b/numpy/core/numeric.py#L1800-L1868 So, I am not sure if these functions exists somewhere in Julia and if there is a need, then where should they be added?

[rdeits]

I'm not sure about the other two, but for something like numpy.indices, we would generally use Julia's CartesianIndices:

julia> CartesianIndices((2,3))
2×3 CartesianIndices{2,Tuple{Base.OneTo{Int64},Base.OneTo{Int64}}}:
 CartesianIndex(1, 1)  CartesianIndex(1, 2)  CartesianIndex(1, 3)
 CartesianIndex(2, 1)  CartesianIndex(2, 2)  CartesianIndex(2, 3)

Note that this returns a slightly different structure than numpy.indices, as CartesianIndices gives you paired (x, y) indices rather than separate arrays of x and y indices.

[timholy]

It would be fine to add. Of course we have edge-detection, but it's not really intended for labeled images. I would probably add it to connected.jl in Images.jl.

As far as connectivity structures goes, check out the docstring and implementation of label_components. This is very old code and hard for a newbiew to understand (it could use some refreshing), but the gist is that we can compile specialized methods for each given connectivity pattern---the stencil function only checks the ones that have been flagged as true. If this is important and desirable, we should consider a more serious revamp of this code to allow it to be generalized.

[rand5]

Speaking of labeled components and connectivity, I think that for people transitioning from MatLab, equivalents to bwperim, regionprops, and imfill would be highly desirable.

[aquatiko]

@timholy I tried implementing the find_boundaries function and this seems to be the result: Original Image: Produced Image:

I used the algorithm where a pixel will be considered a boundary if it's adjacent labels(based on the value of connectivity used in label_components) are not all same. Also we can use this to mark the original image with boundaries with a user given colour. So, should I apply this or is there a better way to do it?

[timholy]

Sounds good to me! I can review it when you push the code.

[mkitti]

https://github.com/JuliaImages/ImageMorphology.jl/releases/tag/v0.3.1 now has the isboundary function. Can this issue be closed?