Some approaches to improve tree-based nearest neighbour search algorithms