虹膜识别算法的DSP移植及其优化

DSP Transplantation and Optimization of Iris Recognition Algorithm

为构建基于DSP(Digital Signal Processor)的虹膜识别系统,需要依据所采用的DSP硬件系统,将现有的虹膜识别算法移植到DSP的开发环境中。本文详细的阐述了虹膜识别算法的DSP移植过程中的关键点,解决了在程序移植过程中所遇到的与硬件系统相关的一系列问题,特别是针对在DSP存储空间有限的状态下存储器和虹膜代码的优化问题。完成了虹膜识别程序的DSP移植,进而实现了基于TMS320DM642的虹膜识别系统。移植后的虹膜识别系统采用DSP/BIOS内核,使用多任务调度,提高了程序执行效率。

Building a iris recognition system based on DSP needs transplanting the existing iris recognition algorithms to the DSP development environment according to the DSP hardware system which is used. The article detailedly elaborated the key point in the process of tranplanting the iris recognition algorithm to DSP, solved a series of problems related to the hardware system encountered during transplantation, especially the iris code and memory optimization problems when the DSP storage space was limited. Having transplanted the iris recognition program to DSP realize the iris recognition system, which is based on TMS320DM642.After transplantation,the iris recognition system uses DSP/BIOS kernel with multi-task scheduling,which enhance the program implementation efficiency.