Object three-dimensional point cloud registration algorithm based on local point pair features

Abstract

The invention discloses an object three-dimensional point cloud registration algorithm based on local point pair features. The algorithm is specifically implemented according to the following steps: step 1, obtaining three-dimensional point cloud data of an object at a plurality of different angles through a depth camera; 2, preprocessing the point cloud data obtained in the step 1; step 3, performing key point extraction on the point cloud obtained in the step 2 to obtain a relatively protruding or recessed point set on the surface of the object; 4, performing feature description on the key point set obtained in the step 3 by using local point pair features to obtain a feature description matrix; step 5, performing feature matching on the feature matrix obtained in the step 4, and calculating a rotation translation matrix by using a matched point pair relationship; and step 6, rotating and translating the point clouds of a plurality of different angles of the object to the same coordinate system by using the rotation and translation matrix obtained in the step 5, and completing registration. The object three-dimensional point cloud registration algorithm based on the local point pair features has the advantages of being high in robustness, high in registration precision and high in operation speed.

Description

technical field [0001] The invention belongs to the technical field of three-dimensional reconstruction of object virtual space, and in particular relates to an object three-dimensional point cloud registration algorithm based on local point pair features. Background technique [0002] Point cloud registration technology is a basic research in the 3D reconstruction technology of object virtual space. Specifically, point cloud registration is the process of registering multiple point clouds from different angles generated by 3D sensors to form a complete 3D model point cloud of the object. With the development of point cloud registration technology and computer vision technology, point cloud registration technology is widely used in 3D reconstruction, target recognition and positioning, robot navigation, shape detection and other fields. [0003] Point cloud registration technology is currently mainly divided into three types: registration algorithm based on feature matching,...

AI Technical Summary

Problems solved by technology

However, because the sensor often contains various noises when acquiring the object point cloud, and the shape of the object point cloud is different, the existing registration algorithm based on feature matching has low accuracy and poor robustness. Therefore, a robust registration algorithm is proposed It is the key to achieve high-quality point cloud registration

Images