Abstract
This paper proposes a method of registering point clouds using 2D images, 3D point clouds, and their correspondences in order to provide appropriate initial conditions for 3D fine registration algorithms such as Iterative Closest Point. Many commercially available optical 3D scanners capture both 3D point clouds and 2D images, and their correspondences can be obtained using camera calibration information. The proposed method registers 3D source data (moving) to 3D reference data (fixed) in an iterative manner, with each iteration consisting of three steps: (1) finding image correspondences in the source and reference images, (2) transforming the source data using the corresponding 3D points, (3) generating a virtual image of the source data in the transformed coordinates. The above steps are repeated until the source data approaches suitable initial conditions for fine registration. The proposed method has been tested on various objects, including mechanical parts, animals, and cultural items.
| Original language | English |
|---|---|
| Pages (from-to) | 1221-1229 |
| Number of pages | 9 |
| Journal | International Journal of Precision Engineering and Manufacturing |
| Volume | 18 |
| Issue number | 9 |
| DOIs | |
| Publication status | Published - 2017 Sept 1 |
Bibliographical note
Funding Information:This research was supported by the Technology Innovation Program (10065150, Development of Low-Cost and Small LIDAR System Technology Based on 3D Laser scanning for 360 Real-time Monitoring), funded by the Ministry of Trade, Industry & Energy (MOTIE, Korea) and the Korea Evaluation Institute of Industrial Technology (KEIT, Korea).
Publisher Copyright:
© 2017, Korean Society for Precision Engineering and Springer-Verlag GmbH Germany.
Keywords
- Coarse registration
- Image registration
- Image reprojection
- Point cloud registration
ASJC Scopus subject areas
- Mechanical Engineering
- Industrial and Manufacturing Engineering
- Electrical and Electronic Engineering