MOVING IMAGE DECODER, MOVING IMAGE ENCODER, MOVING IMAGE DECODING METHOD, AND MOVING IMAGE ENCODING METHOD

• Main Authors: IGAI TOMOHIRO, YASUGI MASANOBU, SASAKI EIICHI, AONO TOMOKO
• Format: Patent
• Language: English; Japanese
• Published: 28.02.2024
• Subjects: ELECTRIC COMMUNICATION TECHNIQUE; ELECTRICITY; PICTORIAL COMMUNICATION, e.g. TELEVISION

To solve a problem in which: in a prediction unit that creates an intermediate prediction image from a product of a matrix and a reference image and interpolates the image to create a prediction image, processing of deriving the intermediate prediction image is complicated.SOLUTION: A moving image decoder comprises: a weighting matrix derivation unit that derives a weighting matrix by using an intra-prediction mode and a size variable; a matrix prediction image derivation unit that derives an intermediate prediction image of a predetermined size by using reference pixels of a target block and product-sum operation of the weighting matrix; and a matrix prediction image interpolation unit that derives a prediction image by using the intermediate prediction image. The predetermined size is a size equal to or less than minimum values of a width and a height of a conversion block, and the intermediate prediction image is derived by using a formula: predMip[x][y]=(((ΣmWeight[i][y*predSize+x]*p[i])+oW)>>sW)+pTemp[0].SELECTED DRAWING: Figure 16 【課題】行列と参照画像との積により中間予測画像を生成し、これを補間して予測画像を生成する予測部では、中間予測画像の導出処理が複雑である。【解決手段】イントラ予測モードとサイズ変数とを用いて、重み行列を導出する重み行列導出部と、対象ブロックの参照画素と、重み行列の積和演算とを用いて、所定サイズの中間予測画像を導出する行列予測画像導出部と、上記中間予測画像を用いて予測画像を導出する行列予測画像補間部と、を備え、上記所定サイズは、変換ブロックの幅と高さの最小値以下のサイズであり、上記中間予測画像は、式predMip [x] [y] = (((ΣmWeight [i] [y * predSize + x] * p [i]) + oW) >> sW) + pTemp [0]を用いて導出されることを特徴とする。【選択図】図16