Constrained approximation of rational triangular Bézier surfaces by polynomial triangular Bézier surfaces (Q526712)

The problem of approximation of rational triangular Bézier surfaces by polynomial Bézier surfaces is considered. The basal polynomials of the presented approach comprise bivariate Bernstein polynomials \(B_{\pmb k}^n(\pmb{x})\) and their \(\pmb{\alpha}\)-parameter dual versions \(D_{\pmb k}^n(\pmb{x};\pmb{\alpha})\). A rational triangular Bézier surface \(\mathsf{R}_n(\pmb{x})\) of degree \(n\) is approximated by a polynomial Bézier surface \(\mathsf{P}_m(\pmb{x})\) of degree \(m\) on the standard triangle \(T\). The control points polynomial coefficients \(p_{\pmb k}\) of \(\mathsf{P}_m(\pmb{x})\) attain prescribed values on the \(\pmb{c}\)-parametrized boundary of the control points set and the \(w_{\pmb{\alpha}}\)-weighted distance between \(\mathsf{P}_m\) and \(\mathsf{R}_n\) is minimized. Explicit theoretical formula for the entire set of control points \(p_{\pmb k}\) involves coefficients \(E_{\pmb l}^{\pmb{k}} (\pmb{\alpha},\pmb{c},m)\) and integral quantities \(I_j\). An efficient recursive algorithm is formulated for fast calculation of \(E_{\pmb l}^{\pmb{k}} (\pmb{\alpha},\pmb{c},m)\) and an adaptive quadrature formula utilizing Chebyshev polynomials of the first kind is developed for \(I_j\). Efficiency of the entire algorithm is demonstrated on model examples. The advantage of the presented method is a possibility of preservation of smoothness of two adjacent rational surfaces.