Embodiment
With reference to following application scenarios, the invention will be further described.
Referring to Fig. 1, including finger print acquisition module 1, D/A converter module 2, fingerprint image processing module 3, fingerprint database
Module 4 and result verification module 5, the rear of finger print acquisition module 1 connect the D/A converter module 2, for gathering target
The infrared image and ultraviolet image of fingerprint;The D/A converter module 2 is used for infrared image and ultraviolet image to target fingerprint
Digital-to-analogue conversion is carried out respectively;The fingerprint image processing module 3 is used to carry out the infrared image of target fingerprint with ultraviolet image
Denoising, decomposition and synthesis processing, obtain target fingerprint image;The fingerprint image for the standard that is stored with the fingerprint database module 4
Picture;The result verification module 5 is connected with the fingerprint image processing module 3 and the fingerprint database module 4, for by mesh
The fingerprint image for marking fingerprint image and standard carries out contrast verification, obtains fingerprint authentication result, and result is shown.
Preferably, when the finger print acquisition module is acquired to the infrared image and ultraviolet image of target fingerprint, use
Infrared COMS imaging lens are acquired to the infrared image of target fingerprint, using ultraviolet CCD imaging lens to target fingerprint
Ultraviolet image is acquired, and LED exciter components are connected with front of ultraviolet CCD imaging lens, and rear is connected with ultraviolet image enhancing
Device, finally connects infrared COMS imaging lens, all camera lens light path coaxials and is placed in parallel.
Preferably, the infrared COMS imaging lens and ultraviolet CCD imaging lens are all double-colored integrated structure.
The above embodiment of the present invention, the infrared image of target fingerprint is merged with ultraviolet image, after fusion
Fingerprint image carries out fingerprint recognition, is greatly improved the accuracy of fingerprint recognition.
Preferably, as shown in Fig. 2 the fingerprint image processing module includes fingerprint image preprocessing submodule, fingerprint image
As resolution process submodule and fingerprint image fusion submodule;The fingerprint image preprocessing submodule is by the infrared figure with noise
Picture and ultraviolet image carry out wavelet transform process, obtain corresponding wavelet coefficient, and the wavelet coefficient now obtained includes noiseless
Wavelet coefficient and noise wavelet coefficients, then utilize the infrared image of improved wavelet threshold function pair target fingerprint and ultraviolet
Image carries out denoising, is specially:
(1) obtained wavelet coefficient is carried out by thresholding processing using improved threshold function table, to filter out noise wavelet system
Number, obtain noiseless wavelet coefficient, the improved threshold function table used for:
In formula,Noiseless wavelet coefficient is represented, ψ is that the infrared image with noise and ultraviolet image are carried out at wavelet transformation
The wavelet coefficient obtained after reason, sgn () is sign function, and i is the variable of sign function, and p and q are adjustable parameter, and υ is small echo
Coefficient threshold, ε is that noise criteria is poor;
As p=0 or q=∞, this improved threshold function table is hard threshold function, and as p=1 and q=1, this is improved
Threshold function table is soft-threshold function;
(2) infrared image and ultraviolet figure of the noiseless wavelet coefficient progress to target fingerprint are obtained after being handled using thresholding
As being reconstructed, noiseless infrared image and noiseless ultraviolet image are obtained.
The above embodiment of the present invention, is passed through by improved threshold function table to the infrared image and ultraviolet image of fingerprint image
Band noise wavelet coefficients in the wavelet coefficient obtained after wavelet transform process are filtered out, then using filter out band noise wavelet
Wavelet coefficient after coefficient carries out Image Reconstruction, obtains side noiseless infrared image and ultraviolet image, carries out fingerprint image and locates in advance
Reason filters out the noise in fingerprint image, to obtain high-quality target fingerprint image graph in target fingerprint image co-registration
Picture, and threshold function table after improving can by adjusting p and q value, make improved threshold function table between soft, hard threshold function it
Between, flexibility when enhancing is filtered out to wavelet coefficient.
Preferably, the fingerprint image resolution process submodule is first to passing through the fingerprint image preprocessing resume module
Afterwards, the noiseless infrared image X obtained1With noiseless ultraviolet image X2Using non-downsampling Contourlet conversion (NSCT) point
Resolution process is not carried out, obtains a respective low frequency sub-band coefficient and a series of high-frequency sub-band coefficient, i.e.,With X1LOW (j, k) and X2LOW (j, k) represent noiseless infrared image X1 and noiseless ultraviolet image X2 respectively
Low frequency sub-band coefficient at pixel (j, k) place,WithRepresent noiseless infrared image X1
With noiseless ultraviolet image X2High-frequency sub-band coefficient in n-th of direction of m-th of yardstick at pixel (j, k) place, M is yardstick
Number, m represents m-th of yardstick, and n is n-th of direction, nmFor the direction number under m-th of yardstick;
Then the sub-band coefficients and 4 pixels of surrounding (j, k) place pixel obtained after NSCT carries out resolution process
The subband coefficient values of point are compared, with the work of the self-defined width noise-free picture pixel of liveness calculation formula node-by-node algorithm two
Jerk, utilizes noiseless infrared image X1With noiseless ultraviolet image X2High-frequency sub-band coefficient carry out liveness calculating, obtain picture
The high-frequency sub-band liveness of vegetarian refreshments (j, k), self-defined liveness calculation formula is:
In formula,WithNoiseless infrared image X is represented respectively1With noiseless ultraviolet image X2
In (j, k) place m-th of yardstick of pixel n-th of direction high-frequency sub-band liveness,WithTable
Show noiseless infrared image X1With noiseless ultraviolet image X2High-frequency sub-band system in m-th of yardstick, n-th of direction at (j, k) place
Number, Difference table
Show noiseless infrared image X1With noiseless ultraviolet image X2It is adjacent with the up, down, left and right four directions of (j, k) place pixel
The high-frequency sub-band coefficient in m-th of yardstick, n-th of direction of pixel;
Similarly, high-frequency sub-band coefficient is changed into low frequency sub-band coefficient, using self-defined liveness calculation formula, to two width
The liveness of the low frequency sub-band of noise-free picture is calculated, and obtains noiseless infrared image X1With noiseless ultraviolet image X2
The liveness W of the low frequency sub-band of (j, k) place pixel1 Low(j, k) and W2 Low(j,k)。
The above embodiment of the present invention, noiseless infrared image and nothing are gone out by self-defined liveness calculation formula node-by-node algorithm
The liveness of pixel in noise ultraviolet image, the readability of pixel is reflected with liveness, and the liveness of pixel is got over
High then the point definition is higher, during according to liveness to infrared image and UV Image Fusion, is conducive to choosing and arrives definition
Higher pixel, obtain that quality is higher, more merged comprising minutia after target fingerprint image.
Preferably, the fingerprint image merges submodule first to noiseless infrared image X1With noiseless ultraviolet image X2
Liveness at point (j, k) place is compared, and corresponding liveness comparative result is obtained, according to liveness comparative result, antithetical phrase
Band coefficient carries out value again, obtains new high-frequency sub-band coefficient and new low frequency sub-band coefficient, and new sub-band coefficients are made
For the sub-band coefficients of the target fingerprint image after fusion, it is specially:
In formula, sub-band coefficients of the X (j, k) for the target fingerprint image after fusion behind point (j, k) place again value, X1
(j, k) and X2(j, k) is respectively noiseless infrared image X1With noiseless ultraviolet image X2Sub-band coefficients at point (j, k) place, W1
(j, k) and W2(j, k) is respectively noiseless infrared image X1With noiseless ultraviolet image X2In (j, k) place pixel liveness,
W1(j, k) includes W1 Low(j, k) andW2(j, K) includes W2 Low(j, k) andX1(j, k) includes
X1 Low(j, k) andX2(j, k) includes X2 Low(j, k) andWork as X1(j, k)=X1 LowWhen (j, k),
X2(j, k)=X2 Low(j, k), W1(j, k)=W1 Low(j, k), W2(j, k)=W2 Low(j, k), when
When, To adjust threshold value,It is set as 0.5, X1 Low(j, k) and X2 Low(j, k) represents noiseless infrared image X respectively1It is purple with noiseless
Outer image X2Low frequency sub-band coefficient at pixel (j, k) place,WithRepresent that noiseless is red
Outer image X1With noiseless ultraviolet image X2High-frequency sub-band coefficient in n-th of direction of m-th of yardstick at pixel (j, k) place,
J and K represent the width and height of two width noise-free pictures respectively;
High-frequency sub-band coefficient by the use of new high-frequency sub-band coefficient and new low frequency sub-band coefficient as composograph with it is low
Frequency sub-band coefficients, by NSCT inverse transformation reconstructed images, the target fingerprint image after being merged, and exported.
The above embodiment of the present invention, to noiseless infrared image and noiseless ultraviolet image pixel liveness size ratio
Compared with, according to comparative result, different values are taken to the target fingerprint Image Sub-Band coefficient after fusion, be conducive to noise infrared image and
The details of noiseless ultraviolet image is complementary so that the target fingerprint image after fusion can include the minutia of two images,
When carrying out fingerprint recognition checking, accuracy of identification is greatly improved.
Finally it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than to present invention guarantor
The limitation of scope is protected, although being explained with reference to preferred embodiment to the present invention, one of ordinary skill in the art should
Work as understanding, technical scheme can be modified or equivalent substitution, without departing from the reality of technical solution of the present invention
Matter and scope.