실험계획법을 통한 실내 포인트 클라우드 데이터의 다운샘플링 기반 복셀 크기의 관한 연구

Abstract

Raw point cloud data that is produced from 3D LiDAR scanning of an indoor environment typically includes unnecessary data that requires pre-processing. A common technique to eliminate noise, is utilizing voxel-downsampling method. However, during the process, appropriate voxel sizes is difficult to obtain. Therefore, it is necessary to research a method of identifying appropriate voxel size. In this preliminary study, design of experiments is conducted to visually observe the effect of voxel sizes on point cloud edges. Results shows that voxel sizes, in the voxel downsampling method, has a direct impact on the edge of point cloud data. It was found that for number of points and corresponding file size, voxel size between 0.3 and 0.4 was most effective whilst edge detection showed most effective in dense point clouds.