Desigh And Implementation Of An Embedded Networked Iris Recognition System Based On DaVinci Technology

Iris recognition is one of the most promising technologies in the 21st century. Compared with other biometric methods, iris recognition possesses the advantages of high accuracy, anti-falsification, non-contact etc. Due to the growing demand of the attendance recording, remote monitoring, access control, the embedded networked iris recognition system will be widely applied around us.Firstly, overall design of embedded networked iris recognition system is performed, and the workflow of each subsystem is detailed described. Secondly, the roles of illuminations, camera lens, image sensors and other necessary parts in iris imaging are analyzed, and a low-cost iris camera with the analysis of Purkinje spots is constructed. Thirdly, based on the DM6446-EVM hardware and software platform, the codec, which contains the core iris recognition algorithm, is packaged for the DSP-side. To achieve the functions of database management, user interface, network communication, the application program framework based on multi-thread technology is developed for the ARM-side. Fourthly, the iris image database named HIT_Iris_Version1.1 has been set up, and the accuracy, rapidity and network performances of the embedded networked iris recognition system are evaluated.Following are the main contributions of this dissertation. First of all, a kind of low-cost iris camera is developed, and the iris image evaluation is based on the analysis of Purkinje spots. Besides, the codec of core iris recognition algorithm is packaged for the DSP-side of DM6446. Furthermore, based on DaVinci technology, the multi-thread application program is constructed under the MontaVista Linux operating system, and it has achieved the functions of user interface, database management etc. Finally, the networked iris recognition is realized, which lays the foundation of large-scale biometric identification platform.