Video encoding and decoding method and apparatus using weighted prediction
AssigneeSamsung Electronics Co., Ltd.
InventorsTammy LEE, Woo-jin HAN
Life of this patent
Estimating the term…
Abstract
Provided is a video encoding and decoding method and apparatus. The video encoding method includes selecting a plurality of reference blocks based on a plurality of motion vectors and setting weights assigned to the plurality of reference blocks independently for each of a plurality of regions of a current block in order to predict and encode the current block, thereby providing a new encoding mode allowing accurate prediction of the current block and thus improving the compression rate of video encoding.
Claims15
A video encoding method comprising:
- estimating a plurality of motion vectors for interprediction of a current block;
- selecting a plurality of reference blocks based on the plurality of estimated motion vectors;
- dividing a current block into a plurality of regions and setting weights assigned to the plurality of reference blocks for interprediction independently for each of the regions;
- predicting the current block from the plurality of reference blocks based on the set weights;
- and
- encoding the current block based on a result of the prediction.
A video encoding apparatus comprising:
- a motion estimation unit which estimates a plurality of motion vectors for interprediction of a current block and selects a plurality of reference blocks based on the plurality of estimated motion vectors;
- a motion compensation unit which divides a current block into a plurality of regions, sets weights assigned to the plurality of reference blocks for interprediction independently for each of the regions, and predicts the current block from the plurality of reference blocks based on the set weights, to generate a prediction result;
- and
- an encoding unit which encodes the current block based on the prediction result.
A computer-readable recording medium having recorded thereon a program for executing a video encoding method comprising:
- estimating a plurality of motion vectors for interprediction of a current block;
- selecting a plurality of reference blocks based on the plurality of estimated motion vectors;
- dividing a current block into a plurality of regions and setting weights assigned to the plurality of reference blocks for interprediction independently for each of the regions;
- predicting the current block from the plurality of reference blocks based on the set weights;
- and
- encoding the current block based on a result of the predicting.
Dependencies are inferred from the claim text.