The nearest neighbor queries of moving objects are of important service terms in LBS(Location-Based Services)field.But the uneven distribution of geographical objects greatly deteriorates the searching efficiency.This paper proposed an optimization nearest neighbor searching algorithm based on k-D tree with the largest empty circle constraint of discrete point set.The experiment showed the new algorithm could be better comparing with non-constraint algorithm in searching efficiency,because it effectively reduced the number of visiting nodes also the number of calculating distance.Furthermore,this algorithm could be used in searching k-order neighbor objects of moving objects to reduce the positioning number in the triangle network.