| The rapid development of information technology not only makes the application environment of identity authentication system more complex,but also puts forward higher requirements for the recognition performance of identity authentication system.The identity authentication technology based on human iris features has absolute advantages in the field of identity authentication due to its high uniqueness,high anti-counterfeiting,high accuracy and non-invasion.However,this technology still has a lot of room for expansion in the recognition performance and application scenarios of the algorithm.Multi-modal recognition technology has become a new breakthrough in the field of biometric identification,and it is also a research focus and hotspot in the field of identity authentication.Aiming at the identification of individual identity in complex environment,this paper starts from the disadvantages of iris recognition technology.Firstly,the shortcomings of existing iris recognition algorithms are optimized.Then,the multi-modal fusion methods of binocular iris fusion and iris and face fusion are studied.Finally,based on the above theoretical research,an embedded multi-modal identity authentication system is designed,which provides a solution for the high-precision identification of individual identity in complex environment.The main research contents and innovations of this paper are as follows:(1)Aiming at the disadvantages of iris recognition that it is easy to collect inferior images when acquiring images and the low efficiency of traditional iris localization algorithms,an iris recognition algorithm based on human eye gray and geometric features is proposed.Firstly,the double-measure evaluation method of iris image overall clarity evaluation and effective area evaluation is used to judge whether the quality of the collected iris image is qualified.Secondly,the iris inner and outer edge positioning process is divided into four parts: pupil separation,pupil positioning,outer edge coarse positioning,and outer edge fine positioning.According to the different gray and geometric characteristics of each part,different methods are used to realize the rapid positioning of the inner and outer edges of the iris.Then,the rubber template method is used to expand the located annular iris region into a rectangular region with the same parameters,and the binarization method of adaptive threshold is used to separate the eyelid eyelash noise region.Finally,this paper uses phase analysis method to extract iris features,binary coding method to encode iris features,and realizes matching recognition by comparing the Hamming distance of different iris feature codes.(2)Aiming at the technical bottleneck of the huge limitations of single-modal biometric recognition technology in application,this paper studies the multi-sample fusion of binocular iris based on iris features and the multi-feature fusion method of iris and face.Firstly,for the multisample fusion of binocular iris,this paper proposes a matching layer dynamic weighted fusion method based on iris image effective area evaluation to solve the problem of poor robustness of the traditional method of using fixed weight to fuse binocular iris in the matching layer.According to the evaluation result of iris effective area in iris quality evaluation algorithm,the proportion of fusion weight of binocular iris is adjusted.Secondly,for the multi-feature fusion of iris and face,this paper constructs a face recognition algorithm based on LBP.The matching layer fusion method of weighted addition is used to fuse the two features of iris and face.The experiment is designed to analyze the iris and face recognition algorithm constructed in this paper.Under which weight fusion can achieve the best recognition effect.(3)Aiming at the problem of poor portability of traditional multimodal identification equipment,an embedded multimodal identity authentication system with multiple authentication methods is designed based on the above algorithm theory research.The hardware of the system selects AIO-RK3399 Pro C embedded motherboard as the data processing platform,and works with various peripheral auxiliary hardware devices such as face recognition camera,iris recognition camera,screen display module and ID card reading module.The Linux operating system is deployed on the RK3399 Pro C embedded motherboard,and the multi-modal recognition software is designed and developed on the QT development platform by using C++ language. |
