摘要 |
<P>PROBLEM TO BE SOLVED: To provide a rounding method and a program ensuring convergence of an input signal to 0 by rounding error when the input signal goes 0, even when numerical operation is carried out with accuracy of finite length by an IIR filter. <P>SOLUTION: The rounding method includes sequentially an operation information acquisition step S101 for acquiring operation information I<SB POS="POST">M</SB>about an IIR filter, a selection step S102 for selecting one rounding process corresponding to the operation information I<SB POS="POST">M</SB>out of more than one predetermined rounding processes, and a rounding step S103 for rounding an input signal S<SB POS="POST">I</SB>to the IIR filter by using the rounding process selected at the selection step S102. <P>COPYRIGHT: (C)2012,JPO&INPIT |