Abstract: The neighborhood search is a key technique of large-scale scattered point clouds data processing.It is a basic algorithm for further processing of point clouds.To solve the problem of low efficiency of the neighborhood search for large-scale scattered point clouds,this paper introduces and realizes a fast neighbourhood searching algorithm based on the binary-encoding octree.First,when constructing the octree,each node of the octree is binary encoded with certain rules.In the process,the three dimensions of the space are encoded respectively.Second,when searching the neighbourhood points,the searching level can be decided by the search radius.Last,when implementing neighborhood search,the required node and its neighbour nodes can be located quickly under the given level.The experiments show that our method is high accuracy and speed,and can realize fast neighborhood search of large-scale scattered point clouds.