基于ARM架构的滤波函数优化  

OPTIMIZATION OF FILTER FUNCTIONS BASED ON ARM ARCHITECTURE

作　　者：陈思润 顾乃杰[1,2,3] 苏俊杰 贺爱香[4] Chen Siren;Gu Naijie;Su Junjie;He Aixiang(School of Computer Science and Technology,University of Science and Technology of China,Hefei 230027,Anhui,China;Anhui Key Laboratory of Computing and Communication Software,University of Science and Technology of China,Hefei 230027,Anhui,China;Institute of Advanced Technology,University of Science and Technology of China,Hefei 230027,Anhui,China;Institute of Information Engineering,Anhui XinHua University,Hefei 230088,Anhui,China)

机构地区：[1]中国科学技术大学计算机科学技术学院,安徽合肥230027 [2]中国科学技术大学安徽省计算与通信软件重点实验室,安徽合肥230027 [3]中国科学技术大学先进技术研究院,安徽合肥230027 [4]安徽新华学院信息工程学院,安徽合肥230088

出　　处：《计算机应用与软件》2018年第9期138-143,共6页Computer Applications and Software

摘　　要：随着人们对移动设备图像视觉的追求日益提高,Android设备的图像处理速度难以满足移动客户端海量应用的性能需求。分析并研究ARM架构硬件特性,使用Cortex-A系列支持的SIMD数据级并行计算技术NEON对OpenCV函数库中的滤波函数进行优化。具体介绍中值滤波的优化过程,且在Cortex-A72开发板上进行充分的测试。实验结果表明,中值滤波函数优化效果显著,对比OpenCV源码性能提升了17倍,图像处理模块中的其他滤波函数均有较大的性能提升。With the increasing pursuit of image vision for mobile devices, the image processing efficiency of Android devices is hard to meet the performance requirements of mass applications in mobile client. We analyzed and studied the hardware features of ARM architecture. The filter functions in the OpenCV function library were optimized by using NEON, the SIMD data-level parallel computing technique which is supported by Cortex-A series. We introduced the optimization process of the median filter in detail and fully tested it on the Cortex-A72 development board. Experimental results show that the median filter function has a significant effect in optimization. It improves performance by as much as 17 times compared with OpenCV. And other filter functions of the image processing module also have great performance improvements.

关 键 词：ARM Cortex—A NEON SIMD OPENCV 中值滤波 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]