The International Journal of Advanced Manufacturing Technology

, Volume 32, Issue 5, pp 505–515

Efficient registration for precision inspection of free-form surfaces

Authors

    • NSF Engineering Research Center for Reconfigurable Manufacturing Systems, College Of EngineeringUniversity Of Michigan
  • Jacob Barhak
    • NSF Engineering Research Center for Reconfigurable Manufacturing Systems, College Of EngineeringUniversity Of Michigan
  • Vijay Srivatsan
    • NSF Engineering Research Center for Reconfigurable Manufacturing Systems, College Of EngineeringUniversity Of Michigan
  • Reuven Katz
    • NSF Engineering Research Center for Reconfigurable Manufacturing Systems, College Of EngineeringUniversity Of Michigan
ORIGINAL ARTICLE

DOI: 10.1007/s00170-005-0370-9

Cite this article as:
Zhu, L., Barhak, J., Srivatsan, V. et al. Int J Adv Manuf Technol (2007) 32: 505. doi:10.1007/s00170-005-0370-9

Abstract

Precision inspection of free-form surface is difficult with current industry practices that rely on accurate fixtures. Alternatively, the measurements can be aligned to the part model using a geometry-based registration method, such as the iterative closest point (ICP) method, to achieve a fast and automatic inspection process. This paper discusses various techniques that accelerate the registration process and improve the efficiency of the ICP method. First, the data structures of approximated nearest nodes and topological neighbor facets are combined to speed up the closest point calculation. The closest point calculation is further improved with the cached facets across iteration steps. The registration efficiency can also be enhanced by incorporating signal-to-noise ratio into the transformation of correspondence sets to reduce or remove the noise of outliers. Last, an acceleration method based on linear or quadratic extrapolation is fine-tuned to provide the fast yet robust iteration process. These techniques have been implemented on a four-axis blade inspection machine where no accurate fixture is required. The tests of measurement simulations and inspection case studies indicated that the presented registration method is accurate and efficient.

Keywords

Optical inspectionRegistration between point set and CAD modelICP algorithmApproximated nearest nodeSimulated measurement

Abbreviations

BIM

Blade inspection machine

CAD

Computer aided design

CMM

Coordinate measuring machine

ICP

Iterate closest point

MCS

Measurement coordinate system

PCS

Part coordinate system

RMS

Root mean square

SNR

Signal to noise ratio

Copyright information

© Springer-Verlag London Limited 2006