|
Curve fitting is important in computer - aided design, graphics, pattern recognition, and picture processing. Using the matrix diagonalization method, this thesis gives a new and fast vectorized algorithm for solving the open B - spline curve fitting. The application of our result to solve the closed B - spline curve fitting is also investigated. Given a set of random points, we demon - strate the good performance of our algorithms on the CRAY X - MP EA / 16se supercomputer. Furthermore, we compare the performance of our algorithms (Fortran codes) with the ones using the subroutine 'folrc' (Assembly codes) in MATH LIBRARY ROUTINES and the ones using the subroutine 'sptsv' (Fortran codes) in LAPACK, respectively.
|