Data point cloud downsizing method based on Poisson-disk sampling

Abstract

The invention discloses a data point cloud downsizing method, comprising the following steps of: estimating an initial point cloud normal; carrying out bilateral filtering on the initial point cloud normal; carrying out Poisson-disk sampling on an initial point cloud; estimating a sampling radius through a similar area of an original point cloud and utilizing a neighborhood ball Boolean operation to expand a usable sampling boundary; and on the basis of a point cloud sampling result, respectively supplementing or removing a sampling point from sparse and dense regions to obtain an appointed downsizing quantity. The data point cloud downsizing method disclosed by the invention keeps sharp edge characteristics and boundary data and prevents the sampling point from being locally focused to obtain more balanced distribution; and therefore, the data point cloud downsizing method is very good for application of subsequent high-quality triangularization, drawing based on points, a shape restricting cartoon and the like.

Description

technical field [0001] The invention relates to the field of reverse engineering, in particular to a data point cloud simplification method based on Poisson-disk sampling. Background technique [0002] At present, the most common application mode in reverse engineering is to use scanning equipment based on optical principles to quickly measure the outer surface of parts or molds to form point cloud data, extract geometric features from them, and then reconstruct polygonal or NURBS surfaces. The original scanning point cloud is usually composed of multiple measured single-piece data. There are overlapping areas at the joints, and the data scale is large and unevenly distributed. reconstruction. In response to this problem, the main goal of the point cloud reduction algorithm is to reduce the amount of data and make the sampling points adaptively and evenly distributed according to geometric features. Point cloud reduction usually has two definitions: given the allowable err...

AI Technical Summary

Problems solved by technology

The advantage of the iterative optimal elimination method is that the point cloud distance error before and after sampling is small. The disadvantage is that as the scale of the point cloud increases, due to the large amount of memory and computing consumption required for global sorting and attribute updating, it is not suitable for efficient streamlining of massive point clouds in reverse engineering. , and it is not easy to preserve sharp edge features and boundaries

The hierarchical clustering method has the advantages of high computational efficiency, but its disadvantage is that it is not easy to control the distribution of sampling points and the error

Existing surface resampling methods directly start from the perspective of distribution characteristics, and can obtain the theoretically optimal distribution of sampling points, but because they usually need to solve the distance on the manifold or maintain local dynamic balance, the computing resources they consume are also the largest

Images