| 摘要 |
<p>PURPOSE: To provide a memory unit optimum for picture processings by executing interpolation calculation in real time without increasing a storage capacity. CONSTITUTION: The fineness of interpolation to be obtained is related to the low-order five bits of coordinates X and Y. They are displayed as (x) and (y.) At the time of defining Ka as a contribution (1-x-x+xy,) Kb as the contribution (x-xy,) Kc as the contribution (y-xy) and Kd as the contribution (xy,) Ka, Kb, Kc and Kd are calculated beforehand for all the possible values of (x) and (y) corresponding to the fineness of the interpolation to be obtained. Then, the values are stored in plural ROMs called interpolation matrixes. By addressing a preliminarily programmed memory storing the contributions Ka, Kb, Kc and Kd different by the coordinates (x) and (y,) the value of a function for indicating the contributions is directly obtained. Then, displacement DX and DY for correcting the position of a picture event positioned inside a mesh by a formula is obtained.</p> |
