Abstract: A fast paralleled data access approach for large scale distributed point clouds is proposed in this paper. First,large scale point cloud data is uniformly distributed in the servers,where data is saved according to some data organization. When data in a definite range is required,the servers will immediately access the data by its own organization. This approach is fit for different data organizations. In the experiment,the accelerating ratio reaches 8 when the number of severs is 9 and the required data area is much less than the total data.