FAST K-NEAREST NEIGHBORS SEARCH 1.0

this file you can find a simple but very effective algorithm for Nearest Neighbour Search which I megalomaniacly called the GLTree.You want more? go to the Professional version of GLTreeIt has been designed for uniformly random data, where is the fastest I ever used, but works fine even on sparse ones. If points are too sparse, for example logspace data, search is still performed correctly but speed can degenerate to a brute search algorithm. In these cases a different data structure is needed but for lack of time I havend-deOaot still coded. If query points are close to reference it is very efficient on sparse dataset too.The tree can be build without running any search. The pointer passed to workspace can be used for the above routines. The tree costruction has linear time complexity and it is very fast, so it becomes advantageus against brute search even for a small number of points. Id-deOaod like to point that in GL-Tree searching has linear complexity (on uniform dataset) instead of n*log(n) like in kd-tree.It is possible to choose if return the distances.

Requirements:	No special requirements
Platforms:	Matlab
Keyword:	Advantageus,  Algorithm,  And It,  Are Too,  Be Used,  Brute Search,  Build,  But For,  But Very,  Called,  Can Be,  Close,  Degenerate,  Designed,  Find,  Search,  Speed,  Works
Users rating:	0/10