Method of generating image data

摘要

Search is performed on Intra 16 to obtain a prediction mode leading to a minimum cost, and the minimum cost in Intra 16 and a corresponding prediction mode are stored. Search is performed on Intra 8 to obtain a prediction mode leading to a minimum cost, and then a relationship of magnitude between the stored minimum cost in Intra 16 and the minimum cost in Intra 8 is judged. After that, the minimum cost in Intra 8 and a corresponding prediction mode are stored, and search is performed on Intra 4 to obtain a prediction mode leading to a minimum cost. A relationship of magnitude between cost_intra and the minimum cost in Intra 4 is judged, and Intra 4 is determined as an optimum prediction mode in a case where Intra 4 is smaller.

主权项

1. A method of generating image data for performing image prediction subblock by subblock, the subblock being obtained by segmenting a macroblock in a predetermined encoding size, said image prediction being performed by a plurality of prediction methods defined by a size of said subblock,
the method comprising: a first step of performing interpolation on a pixel of a second subblock serving as an encoding target in a plurality of prediction modes, using a pixel value of a first subblock serving as a reference target; and a second step of obtaining a cost defined by a sum of absolute difference between a pixel value of a first prediction image of said second subblock obtained by said interpolation and a pixel value of said second subblock prior to execution of interpolation, for each of said plurality of prediction modes, to thereby determine a minimum cost among said obtained costs, wherein said first and second steps are performed in a prediction method in which said encoding size is larger to obtain a first prediction mode leading to a first minimum cost, said first and second steps are performed in a following prediction method to determine a second minimum cost using a prediction mode set in advance for said first prediction mode, and said first and second minimum costs are compared with each other to determine an optimum prediction mode through an operation of selecting the smaller minimum cost.