Research On Embedded Iris Recognition Algorithm

In the wake of the progress of science and technology and the improvement of social life quality,people focus more on the security of identity recognition system.Iris recognition technology stands out by virtue of its high stability,high accuracy,hard to forge and non-contact and other advantages.The existing PC platform-based iris recognition system has shortcomings such as high power consumption,large volume,high cost,and inportability.Therefore,this paper designs an iris recognition system based on an embedded platform with low price,high recognition rate,fast running speed and easy to carry,and uses the Raspberry Pi 4th generation B development board and EAIDK-310 development board to realize it.It overcomes the shortcomings of the traditional PC iris recognition system.The main work of the paper includes the following aspects:Aiming at the influence of noise regions such as eyelids and eyelashes on the recognition accuracy,this paper uses an anti-interference fast iris location method suitable for embedded platform operation.The improved Hough circle detection method and Daugman calculus method are combined to locate the inner and outer boundaries of the iris from coarse to fine,and locate the eyelids and eyelashes as input to the feature matching process.This method reduces the blindness of the traditional positioning method search,and can quickly locate the iris boundary with high positioning accuracy.In order to overcome the problem of too many shift comparison times and high time complexity in the traditional Hamming distance method.This paper uses an improved Hamming distance matching method that automatically controls shift registration,which effectively reduces the number of comparisons,ends the matching process early,and improves the real-time performance of feature matching.IRISIA,a binocular iris collector,was used to establish a small iris image database SDU-IDB(Shandong University-Iris Data Base)in a laboratory environment.A total of 115 iris images of the left and right eyes of the undergraduates and graduate students of the School of Microelectronics of Shandong University were collected,a total of 2538 images.The improved iris recognition algorithm is tested based on CASIA-IrisV1,CASIA-IrisV3-Interval and SDU-IDB three iris libraries,using false recognition rate FAR,rejection rate FRR,equal error rate EER and ROC curve and running speed as A measure of the performance of the iris recognition system.The embedded iris recognition system is developed by using the improved iris recognition algorithm,and the embedded Linux operating system is designed to be implanted to the Raspberry Pi 4B development board and EAIDK-310 development board,with the Raspberry Pi 4B and EAIDK-310 board as the core respectively.The board completes the functions of calling and debugging of USB infrared CMOS iris camera,iris recognition process and result display.Through experiments on a large number of iris image samples,it is concluded that the ARM-based embedded iris recognition system built in this paper has realized the characteristics of low development cost,high recognition accuracy,fast and reliable operation,strong anti-interference ability and good mobility.It meets the requirements of iris recognition for the performance and functions of embedded systems and has certain practical application value.