Quadric Surface Direct Fitting Based on Normal Vectors of Random Data

A new method named quadric surface direct fitting based on normal vectors of measured data is proposed to overcome the current quadric surface fitting method's shortcomings.In the current method,the calculations of the initial geometric parameters are imprecise,and the using of it in surface fitting can cause poor efficiency and stability.Then,the performance contrasts of this method with existing methods are presented based on various test models.Finally,the strategy of quadric surface extraction is established based on the proposed surface fitting method,and the extraction of quadric surface from random data is also provided.Both the experiment data and theoretic analysis prove that the proposed method can give initial geometric parameters precisely,thus enhance the capability of surface fitting method and provide new technique of feature extraction and data segmentation in reverse engineering.