A Fast Data Reduction Method for Massive Scattered Point Clouds Based on Slicing

Abstract: This paper puts forward a high-efficiency data reduction method for massive scattered point clouds.  The proposed method is based on slicing technology.  Firstly，the point cloud is subdivided into several layers，then a reference plane is set for each layer，and the points set within each layer are projected to the relevant plane，thus slicing the point cloud for each layer.  Second，points that are important for the shape expression for each sliced point cloud are extracted using a chordal deviation method.  Two key parameter ，layer number and chordal deviation threshold are discussed，concluding is made that the chordal deviation threshold must be smaller than the minimum average chordal deviation for each slice. Reduction results for both the proposed method and the traditional methods are compared in an experiment.  Results show that the proposed method achieves high-quality reduction results in an efficient manner with wet preserved features and details.