Efficient K-nearest neighbors searching algorithms for unorganized cloud points

Methods previously for K-nearest neighbors searching usually inefficient face millions of points. We propose two efficient K-nearest neighbors searching algorithms for unorganized cloud points: parameter adjustable single axes searching algorithm and algorithm based on Morton-code of cloud points. The first algorithm uses direct way to search K-nearest neighbors in 3D space. It sorts points by x, y, z coordinates separately. Then the algorithm search K-nearest neighbors in single axes to reduce the searching space step by step. The second method transforms the data form 3D coordinate field into Morton-code field first and then searches in indirect way. This method shows very high efficiency especially when face big and dense cloud points. Experiments show our two algorithms are more suitable for dealing with static data of big and dense cloud points.