【作者】 李一宏；

【导师】 于哲舟；

【作者基本信息】 吉林大学 ， 计算机应用技术， 2005， 硕士

【摘要】 嵌入式技术已成为应用研究的热点,基于模式识别技术的人像考勤系统日益受到企业管理者的关注,嵌入式人像考勤机是将嵌入式技术与人像识别领域的具体应用相结合的产物,能够解决传统考勤方法带来的诸如代为打卡以及人工监督等许多问题,具有安全性高、使用友好、实时性强、稳定性好等优点,拥有良好的市场前景。本文依据性能、可靠性、开放源代码、低成本以及良好的技术支持等优点选择了Linux 作为嵌入式操作系统,并在此基础上成功地搭建了系统软硬件开发环境。同时,通过研究和比较现有人像检测和识别技术方面的研究成果,选择了一整套算法作为系统软件的核心算法。在上述研究和比较的基础之上,本文使用C 语言编程实现了嵌入式人像考勤机的软件部分,在实际使用中,系统能够正常、稳定的运行,取得了良好的实际应用效果。更多还原

【Abstract】 With the high speed developments of computer technology,semi-conductor technology and electronics technology, embeddedtechnology based on them has been a research focus in application field, andentered into each realm of people living. Actually, embedded system couldbe found at any place, from the electronic watch, mobile telephone, washer,refrigerator, car, to the office teleconference system etc., these of all aredeveloped or reformed by embedded technology. The age of embeddedsystem has already come.This paper discusses deeply in the designing method of embeddedsystem. Through analyzing the requirement of face recognition attendancesystem and the characteristic of embedded system, the developingenvironment is based on embedded Linux operation system. After thesoftware and hardware were configured successfully, the attendance systembased on face recognition technology and embedded system was achieved.The embedded attendance system mainly applies in the usual attendanceinformation management of the enterprise.Face recognition technology is a branch of biometric identification field,which use face physical features, to check a person’s identity, so as to ensuremuch greater security than password and number technology. In a widevariety of face recognition applications, face recognition technology has twokinds of application: recognition and verification. The former is thejudgment" is who", and the later is the judgment " is some person". Theembedded attendance machine belongs to the application of verificationidentity. Currently, attendance system based on face recognition has drawnenterprise managers’attention.At first, the paper analyzes the system’s requirements of software andhardware, according to the characteristics of embedded system. Then themain parts of the system and relationships of them are confirmed.Considering system’s development periods, not having the ability ofdesigning hardware and other factors, the paper decides to use hardwarewhich is on the market. By researching hardware and comparing their speed,cost, real-time performance etc., the hardware platform is decided. It includescamera, Mifare card reader, electronic lock, LCD, compact flash, embeddedboard (NC-670). Embedded operating system is foundation of embeddedapplication software, so it has important status in embedded system. Thispaper compares some famous embedded operating system, and specifiesadvantages and disadvantages using embedded Linux in embedded systemdevelopment. Based on embedded Linux, and according to principles ofpowerful function, easy usage, low cost and better technique support, all kindof development tools are selected, which includes graphical user interface(GUI),hardware drivers, embedded database, programming language andprogramming tools packets. Then this paper analyzes the development modes of embeddedapplication software, and establishes development environment. In thisperiod, hardware environment is built by serial line and Ethernet line. Inorder to support all kinds of necessary function and make kernel smaller,Linux 2.6 kernel is cut. Programming a script file, which can install rootdirectory and boot loader on Target, software environment is built byexecuting the script file. Finally this paper describes implementation of the embedded system,and simply introduces system kernel arithmetic (face recognitiontechnology). In the kernel arithmetic of face recognition, using the result ofresearch ago and test data for reference, Template Match technology isadopted to detecting and locating a face in simple environment. PrincipalComponent Analysis (PCA) is the method for derivation of optimallow-dimensional representation. We select classical Nearest NeighborAlgorithm (NNA) method and Radial Basis Function Neural Network(RBFN) classifier for recognition strategy. The system application softwarestored in compact flash mainly includes two parts: client platform and datamanagement platform. In client platform, the captured image and theinformation from Mifare card reader are processed, and then the resultswhich the data management platform should deal with are stored in database.The information in database is statistic and processing by data managementplatform. The face image will be detected, located, matched and the resultwill be computed according to the threshold both in client program and datamanagement program. Finished all of the work, system can run correctly and steadily andreceive relatively satisfactory experiment outcome under obligationconditions of equality illumination and front face. Now the paper achieves embedded face attendance machine, but thesystem is only a sample. Because it is a production, so it needs a lotsreformation. The system can be integrated with image capturing equipment,radio frequency reader, LCD, embedded motherboard and so on, thus the V-更多还原