Complexity control of HEVC encoders targeting real-time constraints

High Efficiency Video Coding (HEVC) encoders impose several challenges in computing constrained embedded applications, especially under real-time throughput constraints. This paper proposes an adaptive complexity control scheme (CCS) that dynamically adjusts the encoder to the varying computing capabilities of the hardware platform. To design an efficient scheme, an extensive complexity analysis of key HEVC encoding parameters is herein presented. For this analysis, we developed a parameterized complexity model called “arithmetic complexity,” which can be widely applied to any computing platform. Our results demonstrate that the proposed scheme provides time savings ranging from 10 up to 90 % with an average error (between target and effective complexity) of 1.2 %. Our adaptability and control performance analysis show that the scheme rapidly adapts to dynamic set-point adjustments. Compared to state of the art, our complexity control achieves more accurate results and extra features (such as dynamic set-point adjustment) at the cost of minor losses in coding efficiency.