Motion-compensated frame interpolation using patch-based sparseland model

作者：

Highlights：

• We design a patch-based Bidirectional Motion Estimation (BME) to assign a unique Motion Vector (MV) for each patch by using optical-flow ME.

• We construct the sparse model of the intermediate frame according to the MV of each pixel output by the BME module. This sparseland priori is formulated as a Maximum a Posteriori (MAP) estimation problem under the Bayesian framework, and this MAP estimation problem becomes the non-linear sparseland-prior reconstruction model.

摘要

•We design a patch-based Bidirectional Motion Estimation (BME) to assign a unique Motion Vector (MV) for each patch by using optical-flow ME.•We construct the sparse model of the intermediate frame according to the MV of each pixel output by the BME module. This sparseland priori is formulated as a Maximum a Posteriori (MAP) estimation problem under the Bayesian framework, and this MAP estimation problem becomes the non-linear sparseland-prior reconstruction model.

论文关键词：Motion-compensated frame interpolation,Sparseland model,Optical flow,Predictive search,Maximum a posteriori estimation

论文评审过程：Received 25 July 2016, Revised 25 February 2017, Accepted 26 February 2017, Available online 28 February 2017, Version of Record 6 March 2017.

论文官网地址：https://doi.org/10.1016/j.image.2017.02.010