Relative Hausdorff distance

[Nosferican]

Relative Hausdorff distance. It is used in many applications including graphs. A linear time algorithm is available here.

[juliohm]

It is good to check if the definition of the distance is more appropriate to a specific type like a graph or an image. For example, I have implemented the Hausdorff distance for images here as it doesn't make sense for general vectors of numbers: https://github.com/JuliaImages/ImageDistances.jl

Maybe the paper discusses a similar distance applied to graphs?

[Nosferican]

I will check the implementation in ImageDistances.jl (do you have a reference about its use for images?).

For graphs, it just takes two vectors with the degree distribution. The cumulative form is computed and uses linear interpolation for missing values over 1 to the highest number of degrees in either graph domain. The distance is computed based on those two AbstractVector{<:Real}.

[juliohm]

The README.md there has a reference. The original paper where the distance was introduced I can try to find it later.

…

On Fri, Sep 18, 2020, 08:18 José Bayoán Santiago Calderón < ***@***.***> wrote: I will check the implementation in ImageDistances.jl (do you have a reference about its use for images?). For graphs, it just takes two vectors for same length with the number of nodes with that many degrees. The cumulative form is computed and uses linear interpolation for missing values over 1 to the highest number of degrees in either graph domain. The distance is computed based on those two AbstractVector{<:Real}. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#178 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAZQW3NAEAE5S3MDHWRERJDSGM6YDANCNFSM4RR2AMPA> .