Specific embodiment
In conjunction with following application scenarios, the invention will be further described.
Referring to Fig. 1, a kind of smart office access control system based on fingerprint recognition is shown, comprising: fingerprint collecting equipment
1, server 2 and control equipment 3, wherein
The fingerprint collecting equipment 1, for acquiring user fingerprint image, the user fingerprint image and this that will acquire refer to
The line acquisition number of equipment 1 is sent to server 2;
The server 2, including processing module 21, characteristic storage module 22 and instruction sending module 23;
The processing module 21 is obtained and is used for carrying out edge detection and feature extraction processing to received fingerprint image
Family fingerprint characteristic parameter, and will be prestoring with the fingerprint collecting in the user fingerprints characteristic parameter and characteristic storage module 22
Equipment 1 is numbered associated fingerprint characteristic parameter and is matched, and the fingerprint for calculating the user fingerprints characteristic parameter and prestoring is special
The similarity for levying parameter, when similarity is greater than the threshold value of setting, output matching result is correct;
The characteristic storage module 22 is numbered for storing the fingerprint collecting equipment 1 that user prestores, is set with the fingerprint collecting
Standby 1 numbers associated user fingerprints characteristic parameter and numbers corresponding 3 information of control equipment with the fingerprint collecting equipment 1;
Described instruction sending module 23, when matching result for exporting when the processing module 21 is correct, according to connecing
The fingerprint collecting equipment 1 of receipts numbers to the equipment 3 that controls corresponding with the number and sends door open command;
The control equipment 3 refers to for receiving the door open command sent by the server 2 according to the enabling
It enables and opens its corresponding access control system.
Above embodiment of the present invention controls this by the way that fingerprint collecting equipment 1 is arranged in the different gate inhibition of office
Gate inhibition, the fingerprint image that will acquire are uploaded onto the server in 2, are carried out identifying processing to received fingerprint image by server 2, are sentenced
Whether the fingerprint image that breaks with this specifies the associated fingerprint image characteristics parameter of access control system to matching, to control the gate inhibition
The unlatching of system can prestore different matched fingerprint characteristic parameters to different gate inhibitions according to actual needs, make not
Same user can open different gate inhibitions, enhance the controlled level to office door access control system permission, adaptable, intelligence
Change horizontal high;Meanwhile the identity information of different user is distinguished by fingerprint, enabling permission and user can be further strengthened
Security risk caused by consistency, effectively prevention are lost because of IC card in the prior art.
In one embodiment, the server 2 further includes management module 24, the finger image feature for will prestore
Parameter is associated with the number of fingerprint collecting equipment 1, and is stored into the characteristic storage module 22.
Above embodiment of the present invention, by numbering the fingerprint characteristic of designated user and specified fingerprint collecting equipment 1
It is associated, can effectively manage the enabling permission for specifying gate inhibition in designated user and office, by consistency operation, exempt
It needs that the troublesome operation of corresponding gate inhibition could be opened to typing fingerprint in each access control equipment in the prior art, facilitates user
It uses.
In one embodiment, control 3 information of equipment includes controlling IP address, identity information, the number of equipment 3
In it is one or more.
In one embodiment, referring to fig. 2, the processing module 21 further comprises sequentially connected enhancement unit
211, filter unit 212, edge detection unit 213, feature extraction unit 214 and recognition unit 215, wherein
The enhancement unit 211 obtains gray level image for carrying out gray processing processing to received fingerprint image;
The filter unit 212 obtains filtering image for being filtered to the gray level image;
The edge detection process obtains the finger in filtering image for carrying out binary conversion treatment to the filtering image
Line part;
The feature extraction unit 214 carries out feature extraction processing to the fingerprint portion, obtains user fingerprints feature ginseng
Number;
The recognition unit 215, for will be prestored in the user fingerprints characteristic parameter and the characteristic storage module 22
Number associated fingerprint characteristic parameter with the fingerprint collecting equipment 1 and matched, calculate the user fingerprints feature ginseng
The similarity of number and the fingerprint characteristic parameter prestored, when similarity is greater than the threshold value of setting, output matching result is correct.
In one embodiment, the filter unit 212 further comprises sequentially connected first filter unit and
Two filter units, in which:
First filter unit is used to carry out the gray level image to handle except impulsive noise, obtains the first filtering figure
Picture;
Second filter unit is used to carry out first filtering image to handle except Gaussian noise, obtains the second filtering
Image;
The wherein filtering image that second filtering image is exported as the filter unit 212.
Above embodiment of the present invention, due to shadow in impulsive noise and the generally existing office environment of Gaussian noise
Ring acquisition fingerprint image quality, so as to cause the inaccurate problem of user identity identification, thus respectively to fingerprint image into
Row can effectively improve the quality of fingerprint image except impulsive noise and except Gaussian noise is handled, and accurately identify user for system
Fingerprint characteristic parameter is laid a good foundation.
In one embodiment, first filter unit further comprises:
(1) each pixel of the gray level image A is successively traversed, if pixel Aα,βGray value ε (Aα,β)
=0 or ε (Aα,β)=255, then marking the pixel is impulsive noise point, and otherwise marking the pixel is non-pulse noise spot;
(2) impulsive noise point of acquisition is handled respectively, with impulsive noise point Aα,βCentered on point, obtain the pulse
Noise spot Aα,βD × d neighborhood, wherein d=2c+1, c=mic (Δ α, Δ β), wherein Δ α and Δ β respectively indicates the pulse and makes an uproar
The horizontal distance and vertical distance of sound point and the non-pulse noise spot nearest apart from the impulsive noise point, d indicate the ruler of the neighborhood
It is very little;
To impulsive noise point Aα,βIt carries out except noise processed, wherein what is used removes noise function are as follows:
In formula, ε ' (Aα,β) indicate except pixel A after noiseα,βGray value, fα,βIt indicates with impulsive noise point Aα,βFor
Non-pulse noise point set in d × d neighborhood of central point, η (fα,β) indicate set fα,βThe average gray of middle pixel,
eα,β={ g1|g1∈fα,β,|ε(g1)-ε(Aα,β) | < h ' }, indicate impulsive noise point Aα,βCentered in d × d neighborhood for putting with
The gray value differences of central point are less than the non-pulse noise point set of the threshold value h ' of setting;e′α,β={ g2|g2∈fα,β,|ε(g2)-ε
(Aα,β) | >=h ' } indicate impulsive noise point Aα,βCentered on be greater than or wait with the gray value differences of central point in d × d neighborhood for putting
In the non-pulse noise point set of the threshold value h ' of setting, ζ (eα,β) and ζ (e 'α,β) respectively indicate set eα,βWith e 'α,βMiddle pixel
Gray scale intermediate value, τ1, τ2Respectively indicate setting weight factor;
(3) successively by the gray value ε (A of impulsive noise pointα,β) it is updated to ε ' (Aα,β), export the first filtering image.
Above embodiment of the present invention is utilized since gray value is 0 or 255 to impulsive noise in Instrument image
This Characteristics Detection goes out the impulsive noise pixel of Instrument image, and removes noise function using above-mentioned to the impulsive noise pixel
It is handled, impulsive noise pixel neighborhood of a point window size can be adaptively chosen first, according to the non-pulse in neighborhood
Noise pixel point is updated the gray value of noise pixel point, can accurately restore the gray value of impulsive noise pixel,
Improve the accuracy and intelligent level of image processing module 21.
In one embodiment, second filter unit further comprises:
(1) pixel chosen in first filtering image is center pixel, obtains the center pixel respectively
The similarity of point and each pixel in its b × b neighborhood, wherein the similarity function used are as follows:
In formula, I (α0,α(k)) indicate central pixel point α0With neighborhood territory pixel point α(k)Similarity;ε0And ε(k)It respectively indicates
Pixel α0And α(k)Gray value, ψσIndicate the regulatory factor of setting;
(2) according to pixel α in neighborhood(k)With central pixel point α0Similarity construct an ordered setWhereinWherein b2- 1 table
Show the sum of neighborhood territory pixel point, wherein k ∈ [1, b2-1];
(3) it carries out Gaussian noise to the central pixel point according to ordered set to handle, wherein what is used removes noise letter
Number are as follows:
Wherein,
In formula, ε '0Pixel α after expression goes Gaussian noise to handle0Gray value, ε(j)Indicate pixel in ordered set
α(j)Gray value, I (α0,α(j)) indicate pixel α in central pixel point and the ordered set(j)Similarity,Indicate adaptive
The factor should be denoised,Indicate similitude accumulation and,rvIndicate integer set, rv
=1,2 ..., b2-1};
(4) each pixel in the first filtering image is successively traversed, each described pixel is carried out respectively
It states Gaussian noise to handle, obtains each described pixel and go Gaussian noise treated gray value, and according to described each
A pixel goes Gaussian noise treated that gray value forms the second filtering image.
Above embodiment of the present invention further goes Gaussian noise to handle the first filtering image using aforesaid way,
Most suitable neighborhood territory pixel point is adaptively chosen from the neighborhood territory pixel point of target pixel points as standard, to target pixel points
It is filtered, intelligence degree is high, and denoising effect is good.
In one embodiment, second filter unit further comprises:
Wherein, the regulatory factor ψσIt is obtained by following manner:
(1) NSCT is carried out to the first filtering image B to decompose to obtain low-frequency image R, and low-frequency image R is decomposed into mutually
Equitant b × b fritter Bn, calculate each fritter BnStandard deviationWherein b >=3, n ∈ [1, N], N indicate the fritter
Sum;
(2) initializing set γW=G, the number of iterations u=0, wherein G indicates the primary standard difference threshold size of setting;
(3) start iterative process:
(31) it obtains and meets on low-frequency image RLess than γWFritter BnAs weak texture block, and its coordinate is marked to believe
Breath, wherein W indicates the initial threshold of setting;
(32) coordinate information of the weak texture block is mapped to the first corresponding position filtering image B, in the first filtering figure
It is handled as follows as obtaining weak texture quick on B, and to the weak texture block extracted on the first filtering image B:
The weak texture block obtained in first filtering image B is converted into column vector yu, yuIndicate u-th of weak texture quick conversion
At column vector;
The covariance matrix for obtaining weak texture block composition, wherein the covariance matrix function used are as follows:
In formula, ΣyIndicate that the covariance matrix of weak texture block composition, s indicate the sum of weak texture quick, yuIndicate u-th it is weak
The column vector of texture block, Y indicate data set { yuAverage value;
And the standard deviation estimate of the first filtering image B is obtained according to the minimal eigenvalue of the covariance matrixWhereinumin(Σy) indicate covariance matrix ΣyMinimal eigenvalue;
(33) standard deviation estimate is judged: ifThen exportAs regulatory factor ψσ,
And terminate iterative process;IfThen setAbove-mentioned iterative process is repeated, directly
It is greater than the threshold value of setting to the number of iterations u.
Above embodiment of the present invention determines the size of regulatory factor using aforesaid way, can be in unknown Gaussian noise
Standard deviation in the case where, the standard deviation of Gaussian noise point is accurately estimated, and as regulatory factor, can
Help to improve the accuracy of Gaussian noise processing, good wave filtering effect, intelligent level height.
Finally it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than the present invention is protected
The limitation of range is protected, although explaining in detail referring to preferred embodiment to the present invention, those skilled in the art are answered
Work as analysis, it can be with modification or equivalent replacement of the technical solution of the present invention are made, without departing from the reality of technical solution of the present invention
Matter and range.