| 摘要 |
PROBLEM TO BE SOLVED: To provide a signal processor decreasing a table size and computing Y=X<SP>m/n</SP>with a small error. SOLUTION: A scale conversion part 1 generates a scale conversion quantity E determined according to the power number when an input value X is represented with the power number 2, and performs scale conversion of the X by using the scale conversion quantity E to generate a corrected input value G. A coefficient arithmetic part 2 calculates a correction coefficient (k) and a bit shift quantity F determined by scale conversion quantities E, (m), and (n). Through an LUT 3, a corrected output value corresponding to the corrected input value is determined. An arithmetic part 4 corrects the corrected output value H by using the correction coefficient (k) and shifts the bits of the correction result by the bit shift quantity F to find an output value Y. COPYRIGHT: (C)2006,JPO&NCIPI
|
