Feature Recognition and High-Quality Nonuniform Sampling for Spatial Curves

Traditional uniform sampling methods often ignore important features lying on a curve and thus provide unsatisfactory sampling results. To resolve this problem, a high-quality nonuniform sampling method using feature recognition is proposed for spatial curves, which generates a prescribed number of sampled points including feature points and auxiliary points. Firstly, all the approximate locations of local maximum curvature points and maximum torsion points on a curve are obtained through the parabolic interpolation method, and they are chosen after filtering as feature points to better capture intrinsic shape of spatial curves. Then, by defining the characteristic function in a weighted combination of arc length, curvature and torsion, auxiliary points are adaptively selected from the original curve. Compared with three state-of-the-art sampling methods, numerical experiments demonstrate that this method can achieve more high-quality sampling results and has better applicability, thus further improving the B-spline fitting effect for spatial curves. © 2022, Beijing China Science Journal Publishing Co. Ltd. All right reserved.