Method and device for encoding video image, method and device for decoding video image, and program therefor

Abstract

In order to reduce the prediction error energy in motion-compensation inter-screen prediction and improve encoding efficiency, this video image encoding method employing motion compensation with decimal accuracy involves: a step of acquiring a motion vector by performing motion estimation; a step of calculating reference probabilities of decimal pixel positions as indicated by the acquired motion vector; a step of separating interpolation positions, which are the decimal pixel positions, into a plurality of groups on the basis of the calculated reference probabilities; a step of selecting, for each said group of interpolation positions, an interpolation filter coefficient to be used for creating an interpolation prediction image from among a plurality of interpolation filter coefficient candidates; a step of creating, for each said group of interpolation positions, said interpolation prediction image by using the selected interpolation filter coefficient, and performing encoding from said interpolation prediction image by employing said motion compensation with decimal accuracy; and a step of encoding information indicating the grouping of said interpolation positions and information indicating which interpolation filter coefficient is to be used in each said group of interpolation positions.

Description

Technical field [0001] The present invention relates to a technique for improving the performance of an interpolation filter in video coding and improving coding efficiency. [0002] This application claims priority based on Japanese Patent Application No. 2011-141724 for which it applied to Japan on June 27, 2011, and the content is incorporated herein. Background technique [0003] In video coding, in inter-picture prediction (motion compensation) coding in which prediction is performed between different pictures, the frame that has been decoded is referred to, and the motion vector that minimizes the energy of the prediction error is obtained, and the prediction error signal (also Called residual signal) for orthogonal transformation. After that, quantization is carried out, and through entropy encoding, it finally becomes binary data, that is, becomes a bit stream. To improve coding efficiency, reduction of prediction error energy is indispensable, and a prediction method wit...

AI Technical Summary

Problems solved by technology

Although the same theory is also true for a and c, because the results in the horizontal direction will also affect the interpolation in the vertical direction, a and c are transmitted separately without using symmetry

Images