摘要
目标跟踪技术广泛应用于消费电子、工业检测、安防监控及智能交通等领域.由于PC体积大,功耗高,而嵌入式微处理器的资源和处理能力有限,所以本文基于FPGA技术,设计并实现了以粒子滤波算法为核心的目标跟踪系统,以硬件加速技术提升处理性能.通过对影响速度的复杂浮点数运算采用定制指令的硬件加速方式,对于重复性高、运算简单的RGB转HSV等运算,采用IP核硬件加速方式,实现了算法硬件并行化,系统处理速度得到提升.实验表明,通过FPGA技术和硬件加速技术实现的目标跟踪系统能够满足实时性要求,其单位频率的处理性能高于高性能PC机的处理性能的5倍左右.
Target tracking technology has been widely used in many fields,such as consumption electronics,industrial test,safety monitoring and intelligent transportation.Because of large size and high power consumption of PC,and limited resources and processing performance of microprocessor.Therefore,in this paper we designed and implemented a target tracking system based on FPGA platform,making use of particle filtering algorithm and hardware acceleration technology to improve processing performance.The custom instructions are used to improve complex float processing performance.For the processing like RGBtoHSV which is easy and executed frequently,the IP cores are adopted to increase the system speed.The test results demonstrate vide that,by using the FPGA technology and hardware acceleration technology,the system meet real-time tracking requirements,The processing performance of the system based on FPGA is five faster than the one based on high-performance PC.
出处
《小型微型计算机系统》
CSCD
北大核心
2013年第3期580-584,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60970157)资助
关键词
粒子滤波
FPGA
硬件加速
particle filtering
FPGA
hardware acceleration
