| 摘要 |
<p>In the present invention, memory bandwidth is reduced when encoding an image. An image encoding method includes an obtaining step and a generation step. In the obtaining step, a reference image is obtained. In the generation step, interpolation processing in accordance with a motion vector is carried out on the reference image for every luminance component and color difference component, and a prediction image having greater resolution than the reference image is generated. Further, in the generation step, in a case where the size of a block designated as the unit of interpolation processing is smaller than a pre-decided threshold value, a prediction image for a color difference component is generated without carrying out specific interpolation processing indicating interpolation processing in which the number of pixels accessed by a reference image is large.</p> |
