A novel knot selection method for the error-bounded B-spline curve fitting of sampling points in the measuring process

Abstract The B-spline curve has been widely used in the reconstruction of measurement data. The error-bounded sampling points reconstruction can be achieved by the knot addition method (KAM) based B-spline curve fitting. In KAM, the selection pattern of initial knot vector has been associated with the ultimate necessary number of knots. This paper provides a novel initial knots selection method to condense the knot vector required for the error-bounded B-spline curve fitting. The initial knots are determined by the distribution of features which include the chord length (arc length) and bending degree (curvature) contained in the discrete sampling points. Firstly, the sampling points are fitted into an approximate B-spline curve Gs with intensively uniform knot vector to substitute the description of the feature of the sampling points. The feature integral of Gs is built as a monotone increasing function in an analytic form. Then, the initial knots are selected according to the constant increment of the feature integral. After that, an iterative knot insertion (IKI) process starting from the initial knots is introduced to improve the fitting precision, and the ultimate knot vector for the error-bounded B-spline curve fitting is achieved. Lastly, two simulations and the measurement experiment are provided, and the results indicate that the proposed knot selection method can reduce the number of ultimate knots available.