摘要
三维荧光光谱分析是一种面向复杂体系解析的新兴荧光分析技术。因其计算复杂度高,在相关新型便携设备的研制过程中,计算密集部件的硬件结构设计是关键。通过对典型的三线性分解算法进行运行时分析,得出计算密集部件奇异值分解,并根据算法数据和运算特点,研究了它的硬件设计和实现方法。结果表明,使用QR分解转置迭代方法求奇异值在牺牲了部分处理速度的条件下,硬件面积占用小,适合三维荧光光谱分析算法。
Three-dimensional fluorescence spectrometry is a new fluorescence analysis technology.Because of the high computational complexity of algorithms,in the process of development of new portable devices based on three-dimensional fluorescence spectrometry,the structure design of computing-intensive component becomes the key point.This paper gives a run-time analysis to the common three-linear analysis algorithms and discusses the hardware design and implementation method of SVD component. The result shows that,the design method of SVD hardware component based on iterative QRD has a better area efficiency with some acceptable speed sacrifice,and is more appropriate for three-dimensional fluorescence spectrometry analysis algorithms.
出处
《计算机工程与应用》
CSCD
北大核心
2010年第12期60-63,共4页
Computer Engineering and Applications
基金
国家重点基础研究发展规划(973)No.2007CB216404~~
关键词
三维荧光光谱分析
现场可编程门阵列
矩阵运算
奇异值分解
three-dimensional fluorescence spectrometry
Field Programmable Gate Array(FPGA)
matrix calculating
Singular Value Decomposition(SVD)