【摘要】 提出一种适于海量数据的曲线拟合并行算法 .该算法将数据划分到多个处理机上 ,分别进行拟合 ;并通过对相邻两段拟合曲线的边界进行处理 ,可得到C1连续的完整拟合曲线 .更多还原

【Abstract】 A parallel curve fitting algorithm for metadata is proposed. In this algorithm, all the data is divided into a few of processors, and the data on each processor is fitted parallelly with the least square method. The data in the connection of two adjacent fitting curves is fitted according to the functions of the two curves. The whole fitting curve is obtained to reach C 1 standard. This algorithm can solve the problems of the shortage of memory and slow computational speed fitting metadata on PC experiences, thus it has high computational efficiency. The satisfactory fitting result is obtained by realizing this algorithm on Dawn 1?000?A.更多还原