Ball target detection-based automatic registration method for laser point cloud data

Abstract

The invention discloses a ball target detection-based automatic registration method for laser point cloud data. The method comprises the following steps: respectively extracting ball target characteristics from laser point cloud data of two adjacent stations; building a triangle by taking sphere centers of any three non-collinear ball targets as a peak for any station, and taking the area and three inner angles as four rotational invariants of the triangle, so as to build a reference triangle set comprising a plurality of reference triangles by a reference station; building a to-be-registered triangle set comprising a plurality of to-be-registered triangles by a to-be-registered station; matching the reference triangle set with the to-be-registered triangle set; determining a selected same triangle, and calculating the transformation relation of the to-be-registered station and the reference station. By adopting the method disclosed by the invention, automatic detection and automatic registration of the ball target are achieved by a computer automation mode, reduction of expenses of time, manpower and material resources is facilitated, and a non-professional user can obtain a high-quality three-dimensional data model.

Description

technical field [0001] The invention relates to a method for automatic registration of laser point clouds, in particular to a method for automatic registration of laser point cloud data based on spherical target detection. Background technique [0002] Although terrestrial laser has become a relatively mature and widely used technology, the automatic registration of multi-site laser point cloud data (referred to as "automatic registration") is still a relatively popular research direction, which helps to reduce time, manpower, The overhead of material resources, and enables non-professional users to obtain high-quality 3D data models. [0003] In actual operation, the method of target or manual marking is often used for registration. When the target object is difficult to approach, the use of the above methods is limited. Therefore, the automatic registration method between overlapping point clouds based on features becomes industry research hotspots. Currently, feature-ba...

AI Technical Summary

Problems solved by technology

These methods require more complex calculations, and it is difficult to guarantee the efficiency and reliability of the algorithm to identify smaller spheres in larger scenes.

Images