Design And Implementation Of Biometric Recognition Embedded System

Biometrics exploit physical features in human body (such as fingerprint, palmprint, iris, face, etc) or behavioral features (such as voice, signature, gait, etc) to identify a person. Biometrics develop fast both in extent and depth, and it will eventually become a public identification technology.In recent years, palmprint identification has been gradually developed as a new and important modern biometric, which can be utilized to identify a person efficiently. Because of the notable characters such as non-contact, difficult to counterfeit and high reliability, palmprint identification attracts many researchers from all over the world, and obtains some accomplishments. In the support of "863" state-related subjects, this paper researches both the realization of embedded biometrics system and algorithms in palmprint identification, and then integrates them. As the result an integration automatic embedded palmprint identification system is finally designed and partly realized.This paper first analyses and discusses the demands of the task, then puts forward the hardware system design, and determines the feasibility of the scheme. Based on which we choose the embedded development board with ARM microprocessor as the primary hardware platform, then extend which with necessary auxiliary equipments, and achieve the prototype system of biometric collection and identification. The system can implement the elemental functions of image data acquisition, preprocessing, storing and primary palmprint identifying. In order to implement special functions, this paper adopts algorithm of feature extraction based on global texture energy feature and local texture energy feature, in the 1:1 features matching, local texture energy feature match is applied; whereas in the 1:N features matching, layered search scheme is adopted, where we first get a number of candidate templates through rough match of global texture energy feature, and then next the local texture energy feature with these templates, thus we can quickly get the final results of identification. Finally in this paper, the algorithms of palmprint identification is implemented and debugged on a PC version Qt/Embedded development platform and then transplanted to the ARM embedded system board, At last, a whole palmprint identification prototype is constructed.Primary trials are performed and show that the system runs normally and steadily, and reaches the design requirements of performance of speed. We can change system parameters to adjust False Rejection Rate(FRR) and False Acceptance Rate(FAR) to adapt to different environment. In addition, the system satisfies the need of mobility and real-time processing, and reaches the preliminary design goal.