Font Size: a A A

Embedded Iris Recognition System Based On Dsp Hardware

Posted on:2007-07-19Degree:MasterType:Thesis
Country:ChinaCandidate:Y X LuoFull Text:PDF
GTID:2208360185456309Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
Iris recognition is a rising technology in biometric recognition techniques. It's attractive due to its accuracy, stability, uniqueness and non-invasive character. If the world persons' irises get into a database, it would have appeared the false and refuse false possibility to very fewness. So iris recognition is paid attention to more and more domestic and international relevant researchers because of high veracity and good economic benefits.This paper designs a portable embedded iris recognition system based on digital signal processor (DSP). The system consists of 6 parts: Power Manager and Supervisor Circuit, Iris Acquisition part (CMOS image sensor, CPLD and SRAM), Iris Processing part (DSP), Memory (SDRAM and FLASH), Human-Machine Communicating Equipment (LCD and keyboards) and Data Transmission (USB chip) part. In particular, this paper introduces the use of CPLD to acquire iris images and the application of high performance 32 bit floating-point DSP. The detail of the six modules in hardware and software is discussed. Additionally, some experiences during debugging are expressed.Since the work frequency of DSP is 200MHz and there are lots of components whose work frequencies reach more than 100MHZ, the detail analysis of Signal Integrity (SI) is described in this thesis, including designing, emulation and practice.In this paper, we also introduce the design method of real-time embedded system and the debug experience of DSP/BIOS applications. Because our system is an unattached system, how to solidify codes in this system is expatiated in this paper. While, how to optimize the C code and the use of API (Application Programming Interfaces) functions are briefly introduced.All the hardware and software of our system have been debugged and prove to be stable and high-speed. To process an iris image, this system needs only 0.2 second. The EER (Equal Error Rate) is 3.9% in testing the CASIA iris library. This system can store more than 10,000 iris information and provides an USB2.0 port to interface with PC with 20Mbytes/s.
Keywords/Search Tags:Iris Recognition, DSP, CPLD, Embedded Real-Time System, DSP/BIOS
PDF Full Text Request
Related items