基于复杂曲线表示的切比雪夫多项式拟合并行算法  

Chebyshev Multinomial Approximation Parallel Algorithm Based on Complex Curve Representation

作　　者：邓倩妮[1] 陈笠[2] 陆鑫达[1] 何赢潮[1] 

机构地区：[1]上海交通大学计算机科学与工程系,上海200030 [2]上海交通大学机械与动力工程学院,上海200030

出　　处：《上海交通大学学报》2003年第6期906-909,共4页Journal of Shanghai Jiaotong University

基　　金：国家自然科学基金资助项目 (69773 0 14 )

摘　　要：骨科临床造型系统中进行假体再造时 ,要对 CT片上的原始点采用数学逼近法进行优化处理 .常用的数学逼近法中切比雪夫多项式拟和法误差较小 ,对此 ,提出一种基于复杂曲线表示的切比雪夫多项式拟合并行算法 ,并采用两种 Java方案实现并行程序 .实验结果表明 ,与一般的切比雪夫多项式拟合串行算法相比 ,基于复杂曲线表示的切比雪夫多项式拟合并行算法保持了较高的计算精度 。In the process of re sculpting artificial bone in clinic orthopaedics sculpting system, the sampling data from CT scan image must be optimized and smoothed with mathematical approximating methods. In generic mathematical approximation methods Chebyshev multinomial approximation algorithm has better effect on reducing approximation error. This paper brought forward a parallel Chebyshev multinomial approximation algorithm based on a unique representation of complex curve, and implemented the parallel program in two Java methods. The experiment results show that the parallel Chebyshev multinomial approximation algorithm based on complex curve representation get the significant speedup and better calculating precision compared with normal sequential Chebyshev multinomial approximation algorithm.

关 键 词：并行计算 数据拟合 切比雪夫多项式 多线程 Java远程方法调用 

分 类 号：TP338.6[自动化与计算机技术—计算机系统结构]