Design And Implementation Of Passport Verification System Based On Android Platform

The passport verification system can realize the verification of passport and holder,which is widely used in security inspection,transportation,accommodation and other fields.The passport verification system based on the Android platform can meet the needs of miniaturized,portable,and modular verification of passport and holder information on specific scene.In the process of the passport verification,the low quality of collected multispectral images and unclear anti-counterfeiting feature usually hinder the subsequent identification.In addition,due to the computing power and storage of the Android terminal,a lightweight face verification algorithm is required to achieve the consistency check of passports and holder,which ensures the efficiency of verification.In this paper,we focused on the correction of passport multi-spectral images and the identity verification of passports based on the Android platform,and designed and implemented a software system.First,aiming at the correction of passport multi-spectral images,we proposed an algorithm based on the combination of fractional differential and multi-scale Retinex.It enhanced the V component of the multi-spectral images of the passport in the HSV color space with fractional differential texture to make the edge of character in the infrared image be more legible,and the UV image anti-counterfeiting feature texture be more clear.Then we used the improved multi-scale Retinex(MSR)to enhance the brightness of the V component of the UV image of passport,and it adaptively adjusted the S component,which made the fluorescence anti-counterfeiting feature in UV image be brighter.Further,to solve the problem of consistency verification of the passport and holder based on the Android platform,we proposed a face verification algorithm based on the improved lightweight convolutional neural network Mobile Face Nets.It implemented face detection and alignment based on MTCNN.We proposed reduced Mobile Face Nets-1 for face verification,which reduced model redundancy.We introduced Arc Face loss function to train the network model,which increased the classification boundary interval and improved the efficiency of face verification.Finally,we designed and implemented the passport verification system with interface interaction on the Android platform which realized the passport multispectral image acquisition and correction,passport and holderverification,and information exchange with other platforms.We have conducted functional and performance tests on the system.The test results show that the multi-spectral image correction algorithm is superior to the classic image enhancement methods.Compared with the existing algorithms,the proposed face verification algorithm has the advantages of light weight fast speed and high accuracy.The system is able to exchange information with other platforms,which plays an auxiliary role for expanding the functions of other platforms.After the long time test,it finds that the stability of the system is promising.