AlphaShapes.jl Documentation

Overview of AlphaShapes.jl

Alpha Shapes

An $\alpha$-shape is closely related to the Delaunay triangulation and Convex Hull of a set of points (in any dimension).

The $\alpha$-shape is constructed for a parameter $\alpha$ by testing each Delaunay simplex to determine if it's circumsphere has a radius less than $\alpha$. If it does then the simplex is kept, otherwise the simplex is removed. This modified triangulation is the $\alpha$-shape.

Choosing $\alpha$ is done in many ways. AlphaShapes.jl optimises $\alpha$ to find the triangulation with the smallest volume which includes all input points as a vertex to at least on simplex.

A mathematical reference

Functionality

AlphaShapes.jl represents an $\alpha$-shape as a function of an array, Array{Float64,2}, of points. So all user functionality comes through this.

AlphaShapes.AlphaShape — Function

AlphaShape(X::Array{Float64,2};α::Union{Nothing,Float64}=nothing,
    search::Tuple{Float64,Float64}=(0.0, 10.0),
    MaxSteps::Int=100)::Array{Float64,3}

Find the alpha shape corresponding to the 2D array of points X: [npoints,2], and the α value α.

If α == nothing then a search over the range of values search is done to find the best value. The optimisation objective is the alpha shape area (minimise) subject to all points in X being included in the alpha shape triangulation.

source

Utils

The volume of one simplex can be found using

AlphaShapes.SimplexVolume — Function

SimplexVolume

Calculate the volume of a simplex using the Cayley Menger matrix

https://westy31.home.xs4all.nl/Circumsphere/ncircumsphere.htm

source

For the whole $\alpha$-shape the volume can be found using

AlphaShapes.AlphaShapeVolume — Function

AlphaShapeVolume(A::Array{Float64,3})

return the sum of volumes of all simplices in the alpha shapes A

source

The Delaunay triangulation is used to build the alpha shapes so there is a helper method to build one for any set of points (MiniQhull.jl)

AlphaShapes.GetDelaunayTriangulation — Function

GetDelaunayTriangulation(points::Array{Float64,2})::Array{Float64,3}

Wrap MiniQhull.jl's delaunay to get a delaunay triangualation in any dimension

source

Under the hood

Arbitrary dimensions is handled by using the Cayley-Menger matrix for simplex volume and simplex circumsphere calculation

AlphaShapes.CayleyMenger — Function

    CayleyMenger

Compute the Cayley-Menger matrix from squared
distances dij^2.

https://mathworld.wolfram.com/Cayley-MengerDeterminant.html

For a 2-d triangle the following example would symbolically
represent the matrix for points x1, x2, and x3 where dij^2
is the square Euclidean distance between i and j
Example

julia>:([0    1      1      1
         1    0      d12^2  d13^2
         1    d21^2  0      d23^2
         1    d32^2  d32^2    0])
:([0 1 1 1; 1 0 d12 ^ 2 d13 ^ 2; 1 d21 ^ 2 0 d23 ^ 2; 1 d32 ^ 2 d32 ^ 2 0])

source

AlphaShapes.SimplexCircumSphere — Function

SimplexCircumSphere

Find the centre and radius of the circumsphere of a simplex https://westy31.home.xs4all.nl/Circumsphere/ncircumsphere.htm

source

AlphaShapes.jl Documentation

Overview of AlphaShapes.jl
contents
Index