| 摘要 |
The present invention relates to an image processing apparatus and method in which, in a case where an AIF is employed in motion compensation, the number of calculations for motion vectors is restricted, whereby the load of encoding processing can be reduced. In step S11, a motion vector calculator selects one of the calculated filter coefficients stored in a memory. In step S12, the motion vector calculator calculates a motion vector of each macroblock of an encoding target image by using interpolation pixels which are interpolated with a fraction precision by an AIR using the selected calculated filter coefficient. In step S13, the motion vector calculator calculates the filter coefficient for the encoding target image by using the calculated motion vector so that the prediction error becomes minimum. The calculated filter coefficient is supplied to the memory and is stored therein, and is used for calculating the motion vector for the subsequent image. The present invention can be applied to an image encoding apparatus. |
