Some Improvements in Tree Based Nearest Neighbour Search Algorithms
Nearest neighbour search is one of the most simple and used technique in Pattern Recognition. In this paper we are interested on tree based algorithms that only make use of the metric properties of the space. One of the most known and refereed method in this class was proposed by Fukunaga and Narendra in the 70’s.
This algorithm uses a tree that is traversed on search time and uses some elimination rules to avoid the full exploration of the tree.
This paper proposes two main contributions: two new ways for constructing the tree and two new elimination rules. As shown in the experiment section, both techniques reduce significantly the number of distance computations.
KeywordsNearest Neighbour Metric Spaces Elimination rule Pattern Recognition
- 1.Dasarathy, B.V.: Nearest Neighbour(NN) Norms: NN Pattern Classification Techniques. IEEE Computer Society Press, Los Alamitos (1991)Google Scholar
- 2.Chen, Y.S., Hung, Y.P., Fuh, C.S.: Fast Algorithm for Nearest Neighbour Search Based on a Lower Bound Tree. In: Proceedings of the 8th International Conference on Computer Vision, Vancouver, Canada, vol. 1, pp. 446–453 (2001)Google Scholar
- 8.Alinat, P.: Periodic progress report 4, ROARS project ESPRIT II – Number 5516. Thomson Technical Report TS ASM 93/S/EGS/NC/079 (1993)Google Scholar