Registration of multiview point clouds for application to ship fabrication

Abstract

Ship size is increasing and demand for high precision of the final product is growing, making accurate production of parts and precise assembly increasingly important. Three dimensional scanners can be used to measure the shape of a product, but multiple scanning is required to obtain the entire shape. There is no general method to merge multiview point clouds because registration results are highly sensitive to the point cloud characteristics. This paper proposes a method for registering point clouds along with target detection and improved overlap computation. The proposed method is applied to registration of multiview point clouds, with examples to show practical performance. It is expected that the proposed method can be used in ship fabrication, reducing cost and human effort. (C) 2017 Elsevier Inc. All rights reserved.