Feature Recognition And Non-uniform Sampling Methods For Spatial Curves

Curve and surface fitting is an important research problem in computer aided design and related fields,and sampling method is very important for fitting results.The current state-ofthe-art uniform sampling methods do not consider the actual shape of the curve and often produce unsatisfactory fitting results.Although there are non-uniform sampling methods for plane curves,they are not suitable for spatial curves.Therefore,the purpose of this thesis is to propose a non-uniform sampling method for spatial curves,and select representative sampled points on curves by virtue of feature recognition,so as to improve the fitting effect of spatial curves.This thesis proposes high-quality non-uniform sampling for spatial curves using feature recognition.Firstly,the parabolic interpolation method and parameter field subtraction method are used to obtain all the approximate locations of local maximum curvature points and maximum torsion points on the spatial curve,and then those local maximum points with relatively small values are filtered out to generate the feature points of the curve.Secondly,by defining the characteristic function in a weighted combination of arc length,curvature and torsion,more auxiliary points are adaptively selected from the original curve.Finally,B-spline curves interpolated at all sampled points are obtained by progressive iteration approximation method as fitting curves of the original spatial curve.Compared with three state-of-the-art sampling methods for spatial curves,the proposed method can achieve more high-quality sampling results,thus further improving the B-spline fitting effect for spatial curves,and has better applicability.