Revision history [back]

Quote from Difference of KDTreeIndexParams vs. KDTreeSingleIndexParams:

Yes, that's exactly it. KDTreeIndex performs approximate NN search, while KDTreeSingleIndex performs exact NN search. The KDTreeSingleIndex is efficient for low dimensional data, for high dimensional data an approximate search algorithm such as the KDTreeIndex will be much faster.

Quote from Difference of KDTreeIndexParams vs. KDTreeSingleIndexParams:

 
Yes, that's exactly it. KDTreeIndex KDTreeIndex
performs approximate NN search, while while
KDTreeSingleIndex performs exact NN NN
search. The KDTreeSingleIndex is is
efficient for low dimensional data, data,
for high dimensional data an an
approximate search algorithm such as as
the KDTreeIndex will be much faster.

Quote from Difference of KDTreeIndexParams vs. KDTreeSingleIndexParams:

Yes, that's exactly it. KDTreeIndex performs approximate NN search, while KDTreeSingleIndex performs exact NN search. The KDTreeSingleIndex is efficient for low dimensional data, for high dimensional data an approximate search algorithm such as the KDTreeIndex will be much faster.

Also from the FLANN manual (flann_manual-1.8.4.pdf):

KDTreeIndexParams When passing an object of this type the index index constructed will consist of a set set of randomized kd-trees which will be searched in in parallel.

KDTreeSingleIndexParams

When passing passing an object of this type type the index will will contain a single single kd-tree optimized for for searching lower lower dimen- sionality data data (for example 3D 3D point clouds)