Motion estimation in video coding using simplified optical flow technique

Abstract EN

We propose an interpolation-free sub-pixel Motion Estimation (ME) technique that particularly aims at providing accurate Motion Vectors (MVs), where conventional sub-pixel interpolation ME algorithms used in video coding are too complex and time consuming.

The proposed algorithm is a combination of block matching algorithms and simplified optical flow, which is Taylor approximation.

The technique does not require any pixel interpolation and it is much faster than conventional ME methods.

Statistical results illustrate that the new technique performs quickly and accurately with a compatible performance with respect to the benchmarking full search algorithm.