public static ClusterSet <T> CalculateClusters <T>(
IRSpatialIndex <RPointInfo <T> > index,
double epsilon,
int minimumPointsPerCluster)
where T : IRPointData
{
var points = index.Search().ToList();
var clusters = new List <Cluster <T> >();
foreach (var p in points)
{
if (p.Visited)
{
continue;
}
p.Visited = true;
var candidates = index.Search(p.Point, epsilon);
if (candidates.Count >= minimumPointsPerCluster)
{
clusters.Add(
BuildCluster(
index,
p,
candidates,
epsilon,
minimumPointsPerCluster));
}
}
return(new ClusterSet <T>
{
Clusters = clusters,
UnclusteredObjects = points
.Where(p => p.Cluster == null)
.Select(p => p.Item)
.ToList(),
});
}
private static Cluster <T> BuildCluster <T>(IRSpatialIndex <RPointInfo <T> > index, RPointInfo <T> point, IReadOnlyList <RPointInfo <T> > neighborhood, double epsilon, int minimumPointsPerCluster)
where T : IRPointData
{
var points = new List <T>()
{
point.Item
};
var cluster = new Cluster <T>()
{
Objects = points
};
point.Cluster = cluster;
var queue = new Queue <RPointInfo <T> >(neighborhood);
while (queue.Any())
{
var newPoint = queue.Dequeue();
if (!newPoint.Visited)
{
newPoint.Visited = true;
var newNeighbors = index.Search(newPoint.Point, epsilon);
if (newNeighbors.Count >= minimumPointsPerCluster)
{
foreach (var p in newNeighbors)
{
queue.Enqueue(p);
}
}
}
if (newPoint.Cluster == null)
{
newPoint.Cluster = cluster;
points.Add(newPoint.Item);
}
}
return(cluster);
}