面向YHFT-M7002平台图像中值滤波算法的优化实现  被引量:2

OPTIMIZATION OF MEDIAN FILTER ALGORITHM FOR YHFT-M7002 PLATFORM

在线阅读下载全文

作  者:王梦园[1,2] 柴晓楠 陈云 商建东[1,2] Wang Mengyuan;Chai Xiaonan;Chen Yun;Shang Jiandong(School of Electronics and Information Engineering,Zhengzhou University,Zhengzhou 450000,Henan,China;Henan Supercomputing Center,Zhengzhou 450000,Henan,China)

机构地区:[1]郑州大学电子与信息工程学院,河南郑州450000 [2]河南省超级计算中心,河南郑州450000

出  处:《计算机应用与软件》2023年第9期205-210,241,共7页Computer Applications and Software

基  金:国家重点研发计划项目(X0008606);郑州大学2018年科研启动基金项目(32210919)。

摘  要:随着FT系列处理器在图像处理领域的广泛应用,目前缺少可以充分发挥FT平台优势的高性能图形图像处理库。针对上述问题,在完成FT-M7002平台中OpenCV2.4.9图像库移植的基础上,提出面向该平台的中值滤波算法优化实现。通过分析FT-M7002的体系结构特点与中值滤波算法特性,使用手工向量化、循环展开、双缓冲等手段进行程序优化,充分利用该平台向量运算单元以及向量寄存器资源,提升该算法的数据级与指令级并行性。测试结果表明,相对于中值滤波算法的串行实现,其优化实现能在保证正确性的基础上获得5~16倍的加速效果。With the wide application of FT series processors in the field of image processing,there is currently a lack of high-performance graphics and image processing libraries that can give full play to the advantages of the FT platform.In response to the above problems,based on the completion of the OpenCV2.4.9 image library transplantation in the FT-M7002 platform,an optimized implementation of the median filtering algorithm for this platform is proposed.By analyzing the characteristics of the FT-M7002's architecture and median filtering algorithm,we used manual vectorization,loop unrolling,double buffering and other methods to optimize the program.The platform's vector operation unit and vector register resources were made full use of to improve the data of the algorithm level and instruction level parallelism.The test results show that compared with the serial implementation of the median filter algorithm,its optimized implementation can achieve 5-16 times the acceleration effect on the basis of ensuring the correctness.

关 键 词:中值滤波 高性能处理器 向量化 

分 类 号:TP3[自动化与计算机技术—计算机科学与技术]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象