private byte[] GetCurrentMeanPoint()
{
if (PointCount != 0)
{
return(ClusteringUtilities.MeanOf(this, PointDim));
}
else
{
return(Centroid.Values);
}
}
// returns the feature and value of the node and the rightmost index corresponding the left child node in the sorted band, if the feature is -1 then it is a leaf node
private (byte, int, int) GetNodeValues(Interval intv)
{
int[] usedBands = ClusteringUtilities.PickFromRangeWithoutReplacement(_random, 0, BandCount, _bandCountPerSplit);
(int band, int rightMostInd) = GetBestSplit(intv, usedBands);
if (band == -1) // could not split the node, it is a leaf
{
return(GetMaxCountTarget(), -1, -1);
}
else
{
return(_points[band][_currentIndices[rightMostInd]], band, rightMostInd);
}
}
/// <summary>
/// Initialize clusters with centroids determined by means of image points.
/// </summary>
/// <param name="polysInPoints">List of List of image indexes. Each list contains image indexes, the means of these points will be the centroid ponits</param>
protected void InitManualClustersPoly(List <List <int> > polysInPoints)
{
_clusters = new Dictionary <uint, Cluster>();
System.Diagnostics.Debug.WriteLine("polysInPoints.Count: " + polysInPoints.Count);
foreach (List <int> poly in polysInPoints)
{
System.Diagnostics.Debug.WriteLine("poly.Count: " + poly.Count);
Centroid centroid = new Centroid(PointDim);
List <byte[]> points = new List <byte[]>();
foreach (int p in poly)
{
points.Add(_points[p]);
}
byte[] sample = ClusteringUtilities.MeanOf(points, PointDim);
centroid.CopyValuesFrom(sample);
_clusters.Add(centroid.ID, new Cluster(_points, _distance, centroid));
}
}
/*private void InitSortingBand(int count)
* {
* _currentBand = new byte[count];
* }*/
private void BootstrapAggregateIndices(int pointCount, int bootstrapPointCount)
{
_currentIndices = ClusteringUtilities.PickFromRangeWithReplacement(_random, 0, pointCount, bootstrapPointCount);
}
public void SetWeightedAverageOf(Centroid ctr1, int weight1, Centroid ctr2, int weight2)
{
Values = ClusteringUtilities.WeightedAverageOf(ctr1.Values, weight1, ctr2.Values, weight2);
}
public void SetMeanOf(IEnumerable <byte[]> points)
{
Values = ClusteringUtilities.MeanOf(points, Dim);
}
public void Subtract(byte[] other)
{
Values = ClusteringUtilities.Subtract(Values, other);
}
public void Add(byte[] other)
{
Values = ClusteringUtilities.Add(Values, other);
}
