Anew K-NN query algorithm based on grid clustering of the neighbor objects

Abstract

K-NN query algorithm is one of the important applications in spatial database, using the previous methods of positioning queries and range queries can not well solve the K-NN query problem, the traditional K-NN search algorithms use measurement distance and pruning strategy to search in the adopted index tree, based on the analysis of the basic concepts of KNN query algorithm, use the fast performance of grid index in querying , apply the clustering algorithm into the K-NN query process, a new K-NN query algorithm based on grid clustering of the neighbor objects is proposed in this paper, the algorithm first will find the former K nearest neighbors by using of the traditional methods, then cluster the non-empty grid cells around the K nearest objects so as to achieve the next area to be queried selectively, the experiments show that the performance of the new algorithm is better than that of the traditional query algorithm which will search in the eight neighbor grid cells around the queried object and expand the query scope layer by layer in the grid division region, it is a new method and has a wide application in practice.