TAnnDecomp 

virtual void Print( // print the tree (for debugging) ANNbool with_pts, // print points as well? std::ostream& out); // output stream virtual void Dump( // dump entire tree ANNbool with_pts, // print points as well? std::ostream& out); // output stream virtual void getStats( // compute tree statistics ANNkdStats& st); // the statistics (modified) };
----------------------------------------------------------------------// Box decomposition tree (bd-tree) // The bd-tree is inherited from a kd-tree. The main difference // in the bd-tree and the kd-tree is a new type of internal node // called a shrinking node (in the kd-tree there is only one type // of internal node, a splitting node). The shrinking node // makes it possible to generate balanced trees in which the // cells have bounded aspect ratio, by allowing the decomposition // to zoom in on regions of dense point concentration. Although // this is a nice idea in theory, few point distributions are so // densely clustered that this is really needed. // ----------------------------------------------------------------------// 

TAnnOrthHalfSpace 

Array types The following array types are of basic interest. A point is just a dimensionless array of coordinates, a point array is a dimensionless array of points. A distance array is a dimensionless array of distances and an index array is a dimensionless array of point indices. The latter two are used when returning the results of k-nearest neighbor queries. 

TAnnOrthRect 

This is record Recognion_SearchTree.TAnnOrthRect. 

TAnnShrinkRule 

the authors' suggested choice 

TANNsplitRule 

the authors' suggestion for best 

TMkNode 

This is record Recognion_SearchTree.TMkNode. 

TPqNode 

This is record Recognion_SearchTree.TPqNode.