A Weighted Low-rank Matrix Approximation Based Template Matching Scheme for Inter-frame Prediction

Abstract

In the field of video coding, inter-frame prediction plays an important role in improving compression efficiency. This task is achieved by finding a predictor for each block such that the residual data can be close to zero as much as possible. For recent video coding standards, motion vectors are required for a decoder to locate the predictors during video reconstruction. Block matching algorithms are usually utilized in the stage of motion estimation to find such motion vectors. For decoder-side inter-frame predictive coding, proper templates are defined and template matching algorithms are used to produce a predictor for each block such that the overhead information to transmit motion vectors can be avoided. However, the conventional criteria of either block matching or template matching algorithms may lead to the generation of worse predictors. To enhance coding efficiency, a weighted low-rank matrix approximation approach for decoder-side inter-frame prediction is proposed in this paper. By finding dominating block candidates as well as their corresponding importance factors, a predictor for each block is formed. Together with mode decision, the coder can switch to a better mode between the motion compensation by block matching and the proposed template matching scheme which can operate at decoder side.