| 摘要 |
The present invention reduces the number of reference pixels to be referenced in order to reduce redundancy between color components while suppressing deterioration in encoding performance. An intra prediction part (20A) that performs intra-frame prediction comprises an adaptive thinning-out luminance reference pixel acquisition part (21A) and an adaptive thinning-out color-difference reference pixel acquisition part (22A). If an input image (a) is an image in a non-YUV420 format and the encoding unit is a smallest CU block, the adaptive thinning-out luminance reference pixel acquisition part (21A): thins out, according to a luminance intra prediction mode, reference pixels surrounding a luminance block corresponding to a block targeted for color-difference prediction; and acquires a pixel value of the thinned-out reference pixels. If an input image (a) is an image in a non-YUV420 format and the encoding unit is a smallest CU block, the adaptive thinning-out color-difference reference pixel acquisition part (22A) thins out, according to the luminance intra prediction mode, reference pixels surrounding the block targeted for color-difference prediction, and acquires a pixel value of the thinned-out reference pixels. |
