Optimal HEVC encoding based on GOP configurations

The emergence of HEVC software implementations allows for several different encoding options. Unfortunately, there is no established method for selecting optimal encoding configurations. The current paper proposes the use of a multi-objective optimization framework for selecting optimal encodings that can be subsequently used for solving constrained optimization problems in quality, bitrate, and encoding time. The proposed optimization framework is used to select optimal configurations from 3,600 possibilities based on GOP configurations and other relevant HEVC software options. We demonstrate our approach using the x265 encoder with examples from the UT LIVE video quality database and a standard 2K video example. The results demonstrate the success of the proposed approach by selecting optimal configurations and eliminating several suboptimal encodings.