Laser point cloud reduction method based on dynamic grid k neighborhood search

Abstract

The invention relates to a laser point cloud reduction method based on dynamic grid k neighborhood search. According to the method, a curvature of a point, an average normal included angle between a point and a neighbor point, and an average distance between a point and a neighbor point are calculated by a k neighbor of point cloud data points; on the basis of the three parameters, a feature determination parameter and a feature threshold are defined, the feature determination parameter and the feature threshold are compared, and extraction and keeping are carried out on the feature points; and then secondary reduction is carried out on non-feature points by using a bounding box method, the reduced point cloud and the feature points are spliced, and thus reduced point cloud data are obtained. Compared with the prior art, the laser point cloud reduction method has the following advantages: the geometric features of the model are kept precisely; generation of a blank area is avoided; thecalculation efficiency is improved effectively; and the method has the great practical value.

Description

technical field [0001] The invention relates to a method for reconstructing a three-dimensional model of an object from laser point cloud data, in particular to a laser point cloud simplification method based on dynamic grid k-neighborhood search. Background technique [0002] The point cloud data acquired by existing 3D laser scanners contains rich details and features, and the data volume is huge, including a large number of redundant data points. If the necessary data reduction and preprocessing are not performed, the model reconstruction process will be seriously affected. Efficiency, too dense data points will also affect the smoothness of the target surface reconstruction, and even the model reconstruction cannot be realized due to the existence of noise points. Therefore, it is very important and practical work to streamline the point cloud data under the premise of retaining the model features. [0003] In recent years, the reduction methods for scattered point clou...

AI Technical Summary

Problems solved by technology

The octree method generally uses a recursive data structure to divide the non-leaf nodes into eight sub-nodes, which consumes a lot of computer memory space

Although the improved self-adaptive octree reduces the storage consumption through one simplification, there are still problems such as the need for numbering between adjacent nodes, which makes the process cumbersome.

Images