Registering, Integrating and Building CAD Models from Range Data

In this paper, we introduce two methods for the registration of multiple range images when a prior estimate of the transformationbetween views is not available and the overlap between range images is relatively small. It consists of 2 stages. First, we nd the initial estimated transformation by extracting and matching 3D space curves from di erent scans of the same object. If no salient features are available on the object, we use ducial marks to nd the initial transformation. This allows us to always nd a satisfactory and even highly accurate transformation independent of the geometry of the object. Second, we apply a modi ed Iterative Closest Points algorithm ICP to improve the accuracy of the registration. We de ne a weighted distance function based on surface curvature which can reduce the number of iterations and requires a less accurate initial estimate of the transformation. We demonstrate these methods for the task of integrating disparate range image views to build solid CAD ...

• Type: Conference
• Year: 1998
• Where: ICRA
• Authors: Ruigang Yang, Peter K. Allen
• Keywords: ICRA 1998, Initial Estimated Transformation, Initial Transformation, Range Image, Robotics