摘要
针对软件方式的嵌入式图像处理,ARM Cortex-A系列的处理器集成了基于SIMD指令架构的NEON协处理器引擎,能够实现一条指令同时对多个数据进行并行化处理,可以有效加速多媒体和信号处理算法.本文设计了一种基于FPGA的视频处理系统,充分利用NEON技术对Sobel算法进行优化加速,并与普通软件方法实现算法的方式进行对比,结果显示,采用NEON技术后图像处理加速显著,且视频流畅度好.
Aimed at ARM-based embedded image processing, the ARM Cortex-A series of processors are integrated the NEON coprocessor engine which is based on the SIMD instruction architecture, enabling an instruction to parallelize multiple data simultaneously, which can accelerate multimedia and signaling processing algorithms (such as video codecs, image processing techniques) effectively. In this paper, we design a video processing system based on FPGA, which can make full use of NEON technology to optimize the Sobel algorithm, compared with the way of ordinary software method to implement the algorithm, the results show that NEON technology accelerates image processing significantly, decoded the video more fluent.
作者
陈炎
袁国顺
刘小强
CHEN Yan;YUAN Guo-Shun;LIU Xiao-Qiang(Institute of Microelectronics of China Academy of Sciences, Beijing 100029;University of Chinese Academy of Sciences, Beijing 100049, China)
出处
《微电子学与计算机》
CSCD
北大核心
2018年第7期62-66,共5页
Microelectronics & Computer