A sequential method for spline approximation with variable knots†

In this paper we describe a method for approximating a waveform by a spline. The method is quite efficient, as the data are processed sequentially. The basis of the approach is to view the approximation problem as a question of estimation of a polynomial in noise, with the possibility of abrupt changes in the highest derivative. This allows us to bring several powerful statistical signal processing tools into play. We also present some initial results on the application of our technique to the processing of electrocardiograms, where the knot locations themselves may be some of the most important pieces of diagnostic information.     

Piecewise polynomial approximation of the sixth order with automatic knots detection

Coefficients of a local segment model for piecewise polynomial approximation of the sixth order are evaluated using values of the function and of its first derivative at three knots of the support. The formulas for coefficients of the function’s expansion in degrees of x-x ₀ on a three-point grid are obtained within the recently proposed basic element method. The algorithm for automatic knot detection is developed. Numerical calculations applying quite complicated tests have shown the efficiency of the model with respect to calculation stability, accuracy, and smoothness of approximation.     

Degree elevation of unified and extended spline curves

Unified and extended splines （UE-splines）, which unifl and extend polynomial, trigonometric, and hyperbolic B-splines, inherit most properties of B-splines and have some advantages over B-splines. The interest of this paper is the degree elevation algorithm of UE-spline curves and its geometric meaning. Our main idea is to elevate the degree of UE-spline curves one knot interval by one knot interval. First, we construct a new class of basis functions, called bi-order UE-spline basis flmctions which are defined by the integral definition of splines. Then some important properties of bi-order UE-splines are given, especially for tile transformation formulae of the basis functions before and after inserting a knot into the knot vector. Finally, we prove that the degree elevation of UE-spline curves can be interpreted as a process of corner cutting on the control polygons, just as in the manner of B-splines. This degree elevation algorithm possesses strong geometric intuition.     

Knot selection by boosting techniques

A novel concept for estimating smooth functions by selection techniques based on boosting is developed. It is suggested to put radial basis functions with different spreads at each knot and to perform selection and estimation simultaneously by a componentwise boosting algorithm. The methodology of various other smoothing and knot selection procedures (e.g. stepwise selection) is summarized. They are compared to the proposed approach by extensive simulations for various unidimensional settings, including varying spatial variation and heteroskedasticity, as well as on a real world data example. Finally, an extension of the proposed method to surface fitting is evaluated numerically on both, simulation and real data. The proposed knot selection technique is shown to be a strong competitor to existing methods for knot selection.     

Knot calculation for spline fitting via sparse optimization

Curve fitting with splines is a fundamental problem in computer-aided design and engineering. However, how to choose the number of knots and how to place the knots in spline fitting remain a difficult issue. This paper presents a framework for computing knots (including the number and positions) in curve fitting based on a sparse optimization model. The framework consists of two steps: first, from a dense initial knot vector, a set of active knots is selected at which certain order derivative of the spline is discontinuous by solving a sparse optimization problem; second, we further remove redundant knots and adjust the positions of active knots to obtain the final knot vector. Our experiments show that the approximation spline curve obtained by our approach has less number of knots compared to existing methods. Particularly, when the data points are sampled dense enough from a spline, our algorithm can recover the ground truth knot vector and reproduce the spline.     

调整节点矢量对B样条曲线的影响

本文研究了调整一个节点对B样条曲线产生的影响。调整B样条曲线的一个节点,引起两个节点区间改变。讨论了这两个节点区间上有定义的B样条基函数所发生的变化,以及对B样条曲线产生的影响。研究表明,通过调整一个节点可以方便地调整B样条曲线的形状,为B样条曲线形状调整提供了一种简便有效的方法,这在很大程度上丰富了B样条曲
线形状调整的方法。     

Increasing the usefulness of additive spline models by knot removal

Modern techniques for fitting generalized additive models mostly rely on basis expansions of covariates using a large number of basis functions and penalized estimation of parameters. For example, a mixed model approach is used to fit a model for children’s lung function that allows for non-linear influence of several covariates available in a substantial data set. While the resulting model is expected to have good prediction performance, its handling beyond simple visual presentation is problematic. It is shown how the number basis functions of the underlying B-spline representation can be reduced by knot removal techniques without refitting, while preserving the shape of the fitted functions. The condition for exact knot removal is extended towards approximate knot removal by incorporating the covariance matrix of the initial parameter estimates, resulting in considerable simplification of the model. Covariance matrices for the transformed parameter estimates are provided. It is demonstrated that enforcing the knot removal condition during estimation leads to the difference penalties employed in the P-spline approach for estimation of B-spline coefficients, and therefore provides a further justification for this type of penalty. A final transform to a truncated power basis provides a simple equation for the model. This increases transportability, while retaining properties of the initial fit such as good prediction performance.     

curve fitting by a piecewise cubic polynomial

This paper deals with curve fitting by a piecewise cubic polynomial which is continuous with its first derivative. A knot is inserted successively until a certain criterion is satisfied. Then a suboptimal algorithm is applied to minimize the sum of squares of residuals.     

B-spline backstepping control with derivative matrix estimation and its applications

A B-spline backstepping controller is proposed for a class of multiple-input multiple-output (MIMO) nonlinear systems. The control scheme incorporates the backstepping design technique with a B-spline neural network which is utilized to estimate the system dynamics. The B-spline neural network has the advantage of locally controlling its output behavior compared with other neural networks; therefore, it is very suitable to online estimate the system dynamics by tuning its interior parameters, including control points and knot points. Based on the mean-value theorem, the derivative of B-spline basis functions in relation to parameters can be estimated to online adjust these parameters. In addition, the validity of the proposed scheme is verified through an experiment on a servo motor system which is controlled by the output voltage of the Buck DC-DC converter.     

基于极小化二阶导矢确定节点

构造参数拟合曲线的关键问题之一是为每个数据点指定一个参数值（节点）。提出了一种确定节点的新方法。对于每个数据点,新方法由相邻的三个数据点构造一条二次多项式曲线,二次曲线的节点通过极小化其二阶导矢的平方确定。两个相邻数据点间的节点区间由两条二次曲线确定。为使节点计算公式能有效反映出相邻数据点的变化情况,新方法改进了修正弦长方法并应用于节点计算。新方法是一个局部化方法,因此适合于曲线曲面的交互设计。实验结果说明,新方法比其他节点计算方法有效。     

Construction of the matched multiple knot B-spline wavelets on a bounded interval

In order to solve a boundary value problem using Galerkin's method, the selection of basis functions plays a crucial rule. When the solution of a boundary value problem is not enough smooth or the domain is irregular, multiple knot B-spline wavelets (MKBSWs) with locally compact support are appropriate basis functions. However, to have globally continuous basis functions, a matching across the subdomain interfaces is required. In other words, MKBSWs that are non-zero in the interelement boundaries should be matched. In this paper, we present the primal and dual matched multiple knot B-spline scaling and wavelet functions whose main properties of smoothness and biorthogonality are kept.     

Shape Control and Modification of RationalCubic B-Spline Curves

This paper considers the construction of a rational cubic B-spline curve that willinterpolate a sequence of data points x'+ith specified tangent directions at those points. It is emphasisedthat the constraints are purely geometrical and that the pararnetric tangent magnitudes are notassigned as in many' curl'e manipulation methods. The knot vector is fixed and the unknowns are thecontrol points and x"eightsf in this respect the technique is fundamentally different from otherswhere knot insertion is allowed.First. the theoretical result3 for the uniform rational cubic B-spline are presented. Then. in theplanar case. the effect of changes to the tangent at a single point and the acceptable bounds for thechange are established so that all the weights and tangent magnitUdes remain positive. Finally, aninteractive procedure for controlling the shape of a planar rational cubic B-spline curve is presented.     

Determining Knots by Minimizing Energy

A new method for determining knots to construct polynomial curves is presented. At each data point, a quadric curve which passes three consecutive points is constructed. The knots for constructing the quadric curve are determined by minimizing the internal strain energy, which can be regarded as a function of the angle. The function of the angle is expanded as a Taylor series with two terms, then the two knot intervals between the three consecutive points are defined by linear expression. Between the two consecutive points, there are two knot intervals, and the combination of the two knot intervals is used to define the final knot interval. A comparison of the new method with several existing methods is included.     

B样条乘积计算的区间跳跃算法及应用

利用B样条基函数节点区间的对应关系,首先给出了B样条基函数间的转换矩阵的计算方法,进而给出了计算B样条乘积的区间跳跃算法。该算法仅需计算部分节点区间上的转换矩阵,因此称其为区间跳跃算法。这一方法解决了分段多项式与B样条曲线乘积的计算问题,可应用到B样条曲线的升阶、曲面间光滑拼接等问题中。通过算例验证了该方法计算简捷、易于实现。     

有理多结点样条插值曲线及曲面

鉴于多结点样条曲线（MSIC）是一种点点通过的插值样条曲线,因此在多结点样条插值曲线研究的基础上,给出了有理多结点条插值曲线和有理多结点样条插值曲面的定义,并讨论了有理多结点样条的性质,对有理多结 样条曲线和有理多结点样条曲面的光滑拼接问题进行了讨论,此外,还对有理多结点样条在计算机辅助几何设计中的若干应用问题进行了说明。     

The error in solving a wave equation based on a scheme with weights

The error in approximating a Cauchy problem for a two-dimensional wave equation based on a scheme with weights is studied. The dependence of the approximation error on the time step and weight parameter is considered. With this aim in view, the difference of second-order spatial derivatives in the wave equation is approximated, while the time derivative is preserved continuous; and an analytical solution of a Cauchy problem for a system of ordinary differential equations is obtained as the decomposition in the orthonormal basis consisting of the eigenvectors of the operator of the second difference derivative with respect to the spatial variables. Based on this solution, the errors of the approximation of the wave problem by three-layer difference schemes are studied and the conditions for the stability of a three-layer difference scheme are obtained. It is established that, when simulating the propagation of oscillation processes using difference methods, the oscillation frequency values differ from the real ones and depend on the weight parameter and time step. The optimal values of the weight parameter with which the deviation of the oscillation frequency for the difference scheme is minimal are obtained. The dependences of the approximation error on weight and spatial step are derived. The optimal values of the weight parameter with which the schemes are of the second and fourth order of accuracy with respect to the time step are found.     

The solitary wave solution of the two-dimensional regularized long-wave equation in fluids and plasmas

This paper investigates the solitary wave solutions of the two-dimensional regularized long-wave equation which is arising in the investigation of the Rossby waves in rotating flows and the drift waves in plasmas. The main idea behind the numerical solution is to use a combination of boundary knot method and the analog equation method. The boundary knot method is a meshless boundary-type radial basis function collocation technique. In contrast with the method of fundamental solution, the boundary knot method uses the non-singular general solution instead of the singular fundamental solution to obtain the homogeneous solution. Similar to method of fundamental solution, the radial basis function is employed to approximate the particular solution via the dual reciprocity principle. In the current paper, we applied the idea of analog equation method. According to the analog equation method, the nonlinear governing operator is replaced by an equivalent nonhomogeneous linear one with known fundamental solution and under the same boundary conditions. Furthermore, in order to show the efficiency and accuracy of the proposed method, the present work is compared with finite difference scheme. The new method is analyzed for the local truncation error and the conservation properties. The results of several numerical experiments are given for both the single and double-soliton waves.     

On the detection of peaks and valleys using the local descriptors method

In this correspondence it is demonstrated that the derivative of the angle with respect to an arbitrary independent parameter cannot in general be used to find peaks and valleys of a shape. It is then shown, using concepts from differential geometry, that curvature corresponds to the derivative of the angle with respect to an independent variable if and only if the angle is deflned as a single variable. It is further shown that peaks and valleys detected using curvature as a criterion are insufficient to completely describe a shape.     

三次Hermite插值曲线的细化优化

在给定端点及其切矢方向的条件下,通过在相邻两节点之间插入一个中间节点,研究三次Hermite插值曲线的优化问题．如果以与曲率有关的二阶导数为目标,证明插入节点与不插入节点的情形是一样的,体现三次Hermite插值曲线的一种特性．如果以与挠率有关的三阶导数为目标,给出优化三次Hermite曲线的计算公式,从而提出一种新的曲线构造方法．实例表明了方法的有效性．     

基于改进YOLOv3的木结缺陷检测方法研究

针对木条表面死结和活结缺陷在检测过程中定位困难、平均识别精确度较低、检测速度较慢的问题,在分析木结缺陷特点和改进深度学习YOLOv3模型的基础上,研究其应用于改善木结缺陷检测时的精确度和速度。首先,对活结缺陷图像进行数据扩增,以解决类别不平衡问题。然后,改进k-means++算法,提升木结缺陷目标框的维度聚类效果,得到更合适的初始目标框个数与尺寸;通过缩减YOLOv3中多尺度检测网络、改进损失函数,以减少检测时间和提高目标识别精确度。最后,对木结缺陷进行拼接得出位置坐标。试验结果表明,较改进前YOLOv3算法,mAP值提升7.47%,检测速度提高35%;较Faster R-CNN算法mAP值提升11.68%,检测速度提高约15倍,改进后模型能精确地检测出死结和活结缺陷。因此,在后续研究中,可考虑以YOLOv3算法作为检测木结缺陷模型,进一步改进YOLOv3网络,以提高检测实时性和精确度。