摘要
针对当前图像处理系统存在的处理性能和系统灵活性等问题,提出了一种采用可重构技术和图像并行处理技术实现的图像处理系统。研究了动态可重构技术理论及可重构系统的特点,并且研究了图像并行处理系统的设计及算法实现的方法,分析了目前图像处理系统中存在的问题,利用FPGA(Field)可以多次重复配置的特性,设计了可重构图像并行处理系统。同时,在研究了分布式算法的基础上,实现了图像处理算法。设计了采用多IP核实现图像并行处理系统。系统可以根据计算任务的不同,并同时考虑到并行处理系统负载平衡性,设置不同的计算节点数量,达到了既能够满足系统的需求,又可以节约硬件成本的效果。通过实验,验证了系统的可行性。
In view of the problems of processing performance and system flexibility for current image processing systems, proposed is an image processing system using reconfigurable and image parallel processing technologies. The theory of the dynamic reconfigurable technology and the feature of the reconfigurable system are studied, and the design of parallel image processing system and the methods of algorithm implementation are presented. The problems existing in present image processing systems are analyzed. Making use of the features of multi-reconfiguration for field programmable gate array（FPGA）, the reconfigurable image parallel processing system is designed. On the basis of the research of the distributed algorithm,the image processing algorithm is completed. Designed is the image parallel processing system applying multi IP（Intellectual Property） core. According to the difference of computing tasks and considering the load balancing capability, this system can set different calculating number of nodes to meet the system demand and save the hardware cost.
出处
《半导体光电》
CAS
CSCD
北大核心
2010年第2期291-295,共5页
Semiconductor Optoelectronics
基金
国家"863"计划重点项目(2007AA041501)
长江学者和创新团队发展计划资助(IRT0423)
关键词
可重构计算
动态可重构
图像并行处理
分布式算法
FPGA
reconfigurable computing
dynamic reconfiguration
image parallel processing
distributed arithmetic
FPGA
