Reverse Engineering: Investigation of Optimization Techniques in Point Clouds Registration

Abstract

The Reverse Engineering industries has an increasing need for remanufacturing of worn out spare parts or spare parts with lost technical data like drawings or CAD models, in this project a new hybrid algorithm is proposed for the rigid registration of point clouds obtained from scanning of spare parts by a laser scanner. The hybrid algorithm is dividing the registration process into two stages called the coarse and fine registration, the coarse registration is performed by the Genetic Algorithm (GA) that yield an approximate transformation between the two point clouds, and then inherit its findings to the fine registration that is performed by the Interior Point method to reach the accurate transformation and the successful registration of the two point clouds. In this research two techniques for point clouds registration are applied, tested, evaluated and compared to each other. It was found that using the GA and Interior point method is a good alternative for registering point clouds. The objective of this work is to register two point clouds successfully with minimum error, and high reliability.

Point cloud registration algorithm based on NPFC neighborhood feature descriptor

“…With the rise of 3D imaging technology and the rapid development of 3D scanning technology [1][2], 3D point cloud registration techniques are widely used in many fields, including robot navigation [3], shape recognition [4], and reverse engineering [5]. In the process of scanning and measuring, the object must be scanned several times at different angles due to the influence of object occlusion, environmental factors, or the error of the measuring tool.…”