Research On Fast Acquisition Method Of 3D Fingerprint Based On Structured Light

Biometric identification technology is the main breakthrough direction in the field of information security at present.As an ideal biometric feature,fingerprint has the characteristics of uniqueness,collectability and acceptability,which has been widely used in information security and identity authentication.However,due to the inherent limitations of traditional contact 2D fingerprint collection technology,the development of fingerprint recognition technology has been restricted.With the gradual maturity of 3D reconstruction technology,noncontact collection of 3D fingerprint information has become a new research direction of fingerprint collection technology.In this dissertation,aiming at the requirements of high-speed and high-precision fingerprint collection,the 3D reconstruction method and collection system of fingerprint are deeply studied.The main research work is as follows:(1)To solve the problem of low fringe modulation in high-speed 3D reconstruction by phase shifting profilometry,a fringe modulation enhancing method based on deep learning was proposed.Firstly,the phase error of the low-modulation fringes is analyzed by means of numerical simulation.Secondly,a fringe modulation enhancement network is built.Aiming at the quantitative relationship among the fringe images and the details of fringe image,three modules are introduced in the network,namely,U-shaped structure,residual structure and skip connection module.Finally,performance evaluation experiment of the method is carried out.The experiment shows that the phase error of the simulated data and the real scene data after enhanced by the network is obviously reduced.Compared with the traditional fringe processing algorithm,the method proposed in this dissertation needs less processing time,and it only takes79.5ms to obtain the accurate phase from two phase-shifted fringe images,which could significantly reduce the impact of phase error on the 3D reconstruction accuracy.(2)Aiming at the demand of 3D fingerprint collection,based on the principle of phaseshift profilometry,the fast 3D fingerprint collection technology with the above fringe modulation enhancement method as the core was studied.The hardware structure of projector plus dual cameras and the algorithm module of fingerprint reconstruction were designed,and the fast 3D fingerprint collection system was built.Set up by using the fingerprint acquisition and accuracy verification of experiments,the experimental results show that using the presented algorithm and hardware structure,just under 2000 ?s camera exposure time gathering two phase shift fringes,can obtain accurate phase information,and can realize high precision fingerprint,3D reconstruction of 3D measurement of the mean square error of standard parts within 0.026 mm.The results show that the 3D fingerprint collection system designed in this dissertation has high accuracy and high speed,which can obtain high quality 3D fingerprint data in a very short time,and has the potential to be applied in the actual commercial system.