摘要

A novel method for in-process registration of 3D point clouds scanned from different views is presented. A miniature attitude sensor, which can output its pitch, roll and yaw angles in real-time, is mounted on the scanner. The relative pose between the attitude sensor and the scanner is calibrated in advance by a simple yet effective algorithm. When the scanner is moved from one standpoint to another in a measuring process, the real-time readings of the attitude sensor is utilized to compute the rotation movement of the scanner. After applying the rotation transformation to the current point dataset, the translation movement is efficiently determined by exploiting the normal vector constraint between the correspondence points. The rigid transformation obtained fully automatically can serve as a qualified initial estimate for further fine registration. Experiments demonstrate the applicability of the proposed method.

全文