Abstract:

The point cloud data of industrial components obtained by a 3D laser scanner has high accuracy to represent the object, but it contains a large number of redundant points.On condition that the geometric characteristics of the measured objects are ensured, simplification of the point cloud data is capable of increasing calculation speed, reducing storage space, and highlighting modeling characteristics. In this paper, a new algorithm based on curved surface variation for the point clouds of industrial components is proposed. By calculating the curved surface variations of points, the point cloud is divided into three zones with different characteristics. Weighting values are set for the point zones. The threshold of approximate characteristic points is defined using the curved surface variations of points. Calculate the simplification rates of those points less than the threshold according to its characteristic zones. Then define the distance threshold based on the simplification rate to finish the simplification. The three groups of data for bunny point cloud, point cloud of a box, and point cloud of an industrial part are simplified using the simplification technique introduced in this paper, i.e. curved surface variation-based simplification technique. In respect of extent, speed, and accuracy of simplification, comparison is made between the curved surface variation-based simplification technique and the curvature-based simplification technique. The comparison shows that the two techniques have the same simplification accuracy but the former overtakes the latter in terms of simplification speed. In addition, the former does better in preserving geometric characteristics. Owing to addition of boundary protection processing, the former has little impact on the boundary and can hence meet the subsequent modeling requirements.

Key words: point cloud data simplification, industrial components, curved surface variation, curvature, geometric characteristics, boundary protection