Fast Conic Spline Data Fitting of Noise-Free Data Points

In this paper we address the problem of geometric data fitting of ordered data points, considering the case where the system is much less powerful than the modern personal computers, as in real time systems. Our method achieves timing results around 70 mS for 189 data points and requires very little effort from the system. The proposed method can be summarized in the following three steps: first, input data points are divided into a group of subsets, with respect to a variation of the data points in x or y direction. Then in step2, one data point of each subset is detected, as a shoulder point through a linear search. Finally in step3, detected shoulder point of each subset is utilized to generate quadratic rational Bézier of that subset. The proposed method were implemented and experimentally compared to the well-known geometric fitting methods which have been reported to provide the short execution time and best accuracy. The experiments discussed in this paper show that: the proposed method is significantly faster than timing results of previous methods, since the accuracy of fitting results is reasonable.