一种基于FPGA的手术导航定位系统的硬件加速方法

摘要

Optical tracking system is the core component of image-guided surgical navigation, this paper presents several hardware accelerating methods used in a infrared optical tracking system, which was developed by our team, based on field programmable gate array ( FPGA) . The original tracking system has achieved a location accuracy of submillimeter with SoPC as its computation unit , but had an insufficient tracking speed,which was not able to satisfy the practical application. In order to overcome the lack of computation ability on SoPC ,with hardware accelerating methods ,which are floating-point custom instruction , tightly coupled memory ( TCM) and multiprocessor , calculating speed of 3D reconstruction gained a significant promotion of 18 times according to the experiment results. The hardware acceleration and its performance could theoretically support up to 32 markers real-time location maintaining a frame rate of 60fps. This meets the rate requirement of motion tracking in image-guided surgical navigation.%光学定位系统是手术导航设备中的核心部件,本文针对基于FPGA自行研制的红外光学定位跟踪系统设计了几种硬件加速方法.使用SoPC作为计算单元的原有系统能够达到亚毫米级别的定位精度,但是定位跟踪速率不足,无法满足实际应用需求.为克服SoPC计算能力较弱的问题,本文使用浮点数自定义指令集、紧密耦合寄存器 (TCM) 和多核处理等几种硬件加速方法,使红外光学定位跟踪系统的三维重建速度获得了明显的提升,实验数据表明速度提升18倍.硬件加速之后的系统性能可以支持高达32个标志点以60fps帧率进行的实时定位和跟踪,满足手术导航系统对运动跟踪速率的要求.