Hardware-friendly Advanced Motion Vector Predictor Generation for an HEVC Encoder

• Published in: Journal of semiconductor technology and science Vol. 18; no. 6; pp. 737 - 747
• Main Authors: Park, Seung-Won, Rhee, Chae Eun
• Format: Journal Article
• Language: English
• Published: 대한전자공학회 01.12.2018
• Subjects: 전기공학; motion estimation; parallel architecture; High Efficiency Video Coding(HEVC); advanced motion vector predictor
• ISSN: 1598-1657; 2233-4866
• DOI: 10.5573/JSTS.2018.18.6.737

As the demand for high-definition video services has increased in recent years, high efficiency video coding (HEVC) has adopted as the next-generation compression standard, as of 2013. With regard to hardware-based HEVC encoders, there have been many efforts to improve the processing speed of ME through parallel execution. In this case, the advanced motion vector predictor (AMVP) information must be provided quickly for ME. However, for the parallel processing of AMVP to be accelerated, there is a dependency problem during the generation of the AMVP information. The size of the AMVP hardware also increases sharply. In this paper, two schemes are proposed for a rapid and small AMVP generation module. First, the reduced AMVP scheme avoids the data dependency problem while minimizing hardware resources through the use of approximation. Second, the AMVP reuse scheme correct the approximation to follow the standard with a short execution time. The proposed AMVP hardware is implemented and tested in Verilog. Simulation results show that the performance of the AMVP hardware is greatly improved in terms of the speed and resource use, while the degradation of the compression efficiency is negligible. KCI Citation Count: 0