A Differentiated Registration Method for Complex Surfaces
Different error requirements on different surfaces are necessary in many situations. If the registration between the measured dataset and the nominal surface is performed ignoring these differences, an inaccurate result could occur due to the surfaces which have low precision. An improper registration even results in false verdict. So the differentiated registration is of significance in precision engineering. In this paper, a new registration method taking into consideration of the different error requirements is proposed. Firstly surfaces are classified based on different error requirement. Then, appropriate weight factors are iteratively given to these surfaces. Based on the change relationship of error and weight factors, the algorithm controls the error distribution by optimizing the weighted least squares. Our method has the great flexibility in distributing the error, which is suitable for engineering applications. The experiment demonstrates the validity.