Design And Implementation Of Finger Biometric Fusion Recognition Integrated Machine

Compared with other biometric identification systems,the multi-modal feature recognition system based on fingers is relatively easy to collect images,and the three modal characteristics of finger-vein(FV),finger-knuckle-print(FKP)and finger-print(FP)belong to network structure.The expression modes are compatible with each other.Later,the use value of fingers can be explored to the greatest extent,and the problem of insufficient accuracy and stability in finger single-modal recognition can be effectively compensated.Aiming at the shortcomings of contact fingerprints,the poor image quality of the three-modal finger,and the stable acquisition of the Region of Interest(ROI),the software and hardware construction of finger biometric fusion recognition machine are designed and implemented.The specific research content is as follows:(1)An optimization scheme of finger multi-modal acquisition light source is proposed and the hardware construction is completed.The imaging schemes of FV,FKP and FP are designed and light source structure was optimized by Solid Works modeling and Trace Pro ray simulation to improve the imaging quality.According to the imaging plan,the finger positioning and acquisition module are modeled and printed.Circuit integration of the control module is achieved based on functional requirements.(2)A collaborative ROI localization method of finger multi-modal images is proposed.Correcting finger three model images based on postural consistency,then the inflection point detection scheme in finger arch image is improved for delineating FKP and FV ROIs by using corresponding algorithm.According to fingernail root position detection in finger frontal imaging,the FP’s ROI baseline is determined to obtain the non-contact FP ROI.(3)A three-modal feature extraction and fusion method based on Uniform ELBP are proposed.Firstly,the three-modal ROI image is enhanced by Gabor+WLD,and the histogram feature is extracted by feature encoding using Uniform ELBP,and then concatenated into a one-dimensional feature vector as the overall feature information of the finger for matching.Finally,the performance of finger biometric fusion recognition machine is analysised.The results show that the proposed scheme can effectively improve the three modes of finger imaging quality and the robustness of ROI extraction,and the portability and recognition efficiency of machine are also greatly improved.