基于PAAG的OpenVX核心库函数并行化实现  

Parallel implementation of kernels of OpenVX based on PAAG

作　　者：李涛[1] 孙建[2] 王鹏博[2] 

机构地区：[1]西安邮电大学电子工程学院,陕西西安710121 [2]西安邮电大学计算机学院,陕西西安710121

出　　处：《西安邮电大学学报》2015年第2期7-10,共4页Journal of Xi’an University of Posts and Telecommunications

基　　金：国家自然科学基金重大项目(61136002)

摘　　要：针对OpenVX1.0标准中的核心库函数,在新型多态同构阵列处理器平台上实现基于该标准的Kernel函数算法的并行化。以Gaussian Filter、Color Convert、Sobel3×3等kernel函数组成的节点构造图像边缘检测的OpenVX图执行模型,将该执行模型映射到阵列机上实现了并行加速。实验结果证明,PAAG阵列机能实现高复杂度算法的并行实现,并且能获得线性增长的加速比,执行效率更高。In the view of kernel functions of OpenVX1.0standard,this paper presents a parallel implementation method for the kernels of the standard on a new platform,Polymorphic Array Architecture for Graphics and image processing（PAAG）.Meanwhile,parallel implementation of an application,Image Edge Detection based on OpenVX graph mode which is constructed by nodes,Gaussian Filter、Color Convert、Sobel3×3and so on,is realized in this paper.The execution mode is mapped to the array to implement the parallel acceleration.The experimental results show that,the PAAG array can realize parallel implement of highly complex algorithms,and speedup the algorithms linearly and the efficiency of the execution is higher.

关 键 词：图像处理 OpenVX 边缘检测 多态同构阵列机 并行计算 

分 类 号：TP702[自动化与计算机技术—检测技术与自动化装置]