Acceleration of Binning Nearest Neighbor Methods

A new solution method to the Nearest Neighbour Problem is presented. The method is based upon the triangle inequality and works well for small point sets, where traditional solutions are particularly ineffective. Its performance is characterized experimentally and compared with k-d tree and Elias approaches. A hybrid approach is proposed wherein the triangle inequality method is applied to the k-d tree and Elias bin sets. The hybridization is shown to accelerate the k-d tree for large point sets, resulting in 20% improvement in time performance. The space efficiencies for both the k-d tree and Elias methods also improve under the hybrid scheme.