样点自适应补偿(Sample Adaptive Offset,SAO)是高效视频编码(High Efficiency Video Coding,HEVC)的重要组成部分,能有效去除编码图像中的振铃效应,改善编码图像的主观质量.针对当前SAO滤波器处理速度慢和吞吐量较低的问题,基于6×6大小的基本信息统计单元采用亮色度并行处理的四级流水线设计,提出一种基于FPGA的适用于超高清视频(Ultra HD Video,UHD)的SAO硬件滤波器.实验结果表明,在Xilinx Virtex7 FPGA开发平台上,该架构仅占用了 10.5k的查找表和6.2k的寄存器资源,工作主频最高可达250 MHz,支持8K@110 fps的超高清视频流处理.在相同分辨率条件下,相较于现有其它方案在逻辑资源及帧率方面更具优势.
SAO hardware optimization design for UHD video
Sample Adaptive Offset(SAO)is an important component of High Efficiency Video Coding(HEVC),which can effectively remove ringing effects in encoded images and improve the subjective quality of encoded images.In response to the problems of slow processing speed and low throughput of current SAO filters,based on 6 × 6 basic information statistics unit adopts a four stage pipeline design with parallel processing of brightness and chromaticity,and proposes an FPGA based SAO hardware filter suitable for Ultra HD Video(UHD).The experimental results show that on the Xilinx Virtex 7 FPGA development platform,this architecture only occupies 10.5k lookup tables and 6.2k register resources,with a maximum operating frequency of 250 MHz,supporting 8K@110 fps for ultra high definition video streaming processing.Compared to other existing solutions,it has more advantages in logical resources and frame rate under the same resolution conditions.
万方数据
维普
