Manual Preliminary Coarse Alignment of 3D Point Clouds in Virtual Reality

Abstract

The alignment of 3D point clouds consists of coarse alignment and precise alignment. The preliminary coarse alignment must be implemented for point clouds with a significant initial pose difference before time-consuming precise alignment. However, this procedure is normally finished on 2D interfaces manually, which leads to a partial perception of the 3D point clouds. The biased understanding may affect the operation efficiency and alignment accuracy. In this paper, we developed a VR-based prototype for manual preliminary coarse alignment of point clouds. A user study was conducted to compare the efficiency, accuracy, and usability in a controlled alignment task with both the 2D interface and the developed system. The task was graded into three levels based on the complexity of matched points clouds (simple, complex, and incomplete). The result indicated that the prototype system was effective and useful for supporting the preliminary coarse alignment task. It displayed outstanding performance for the coarse alignment of complex and incomplete point clouds.