CN115841685B - Fake fingerprint detection system and method based on composite pixel gradient - Google Patents
Fake fingerprint detection system and method based on composite pixel gradient Download PDFInfo
- Publication number
- CN115841685B CN115841685B CN202310116481.1A CN202310116481A CN115841685B CN 115841685 B CN115841685 B CN 115841685B CN 202310116481 A CN202310116481 A CN 202310116481A CN 115841685 B CN115841685 B CN 115841685B
- Authority
- CN
- China
- Prior art keywords
- composite pixel
- sub
- gradient
- pixel gradient
- composite
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000002131 composite material Substances 0.000 title claims abstract description 203
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000001514 detection method Methods 0.000 title claims abstract description 37
- 238000012549 training Methods 0.000 claims abstract description 19
- 238000012706 support-vector machine Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims description 46
- 238000004364 calculation method Methods 0.000 claims description 19
- 238000000605 extraction Methods 0.000 claims description 17
- 238000013139 quantization Methods 0.000 claims description 14
- 238000012360 testing method Methods 0.000 claims description 13
- 238000003860 storage Methods 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 7
- 238000010586 diagram Methods 0.000 description 9
- 238000012545 processing Methods 0.000 description 8
- 238000004422 calculation algorithm Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 4
- 239000000463 material Substances 0.000 description 4
- 238000005286 illumination Methods 0.000 description 3
- 238000000513 principal component analysis Methods 0.000 description 3
- 238000009826 distribution Methods 0.000 description 2
- 238000003708 edge detection Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000013100 final test Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 108010010803 Gelatin Proteins 0.000 description 1
- 241000282414 Homo sapiens Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 229920000159 gelatin Polymers 0.000 description 1
- 239000008273 gelatin Substances 0.000 description 1
- 235000019322 gelatine Nutrition 0.000 description 1
- 235000011852 gelatine desserts Nutrition 0.000 description 1
- 239000003292 glue Substances 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 239000004816 latex Substances 0.000 description 1
- 229920000126 latex Polymers 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000003825 pressing Methods 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 239000011347 resin Substances 0.000 description 1
- 229920005989 resin Polymers 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000003313 weakening effect Effects 0.000 description 1
- 239000002023 wood Substances 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a fake fingerprint detection system and method based on composite pixel gradients, comprising the steps of preprocessing a fingerprint image set, extracting an interested region of the fingerprint image, constructing composite pixel gradients and composite pixel gradient feature matrixes, reducing dimensions of a plurality of composite pixel gradient feature matrixes, inputting the composite pixel gradient feature matrixes into a support vector machine, obtaining a true and false judging model of the fingerprint image after training, and detecting the tested fingerprint image to improve accuracy of true and false fingerprint judgment.
Description
Technical Field
The invention belongs to the technical field of digital image processing, and particularly relates to a fake fingerprint detection system and method based on a composite pixel gradient.
Background
With the development of related technologies such as artificial intelligence and pattern recognition, the research and development and production technologies of fingerprint sensors are mature, a large number of high-resolution fingerprint images can be captured, and the fingerprint sensors are widely applied to the fields of finance, public security, access control, household registration management and the like. However, as the fingerprint information of human beings is easy to steal and forge, the fingerprint identification system brings a convenient identity authentication mode to people and brings potential safety hazards of identity authentication; in addition, different input angles can bring challenges to fingerprint identification and authentication technology when fingerprints are input, and in view of the texture characteristics of fingerprint images and the conditions of rotation, illumination and other influencing factors when fingerprints are acquired, no effective fingerprint characteristic extraction method is available in the prior art to realize the identification and extraction of composite characteristics.
At present, an image feature extraction machine learning algorithm has a directional gradient histogram feature HOG (Histogram of directional gradients), scale-invariant feature transform (SIFT-invariant feature transform) and the like, wherein the HOG feature extraction method normalizes a color space through Gamma correction, reduces the influence of illumination variation, and also partially reduces the interference of noise. Because the HOG features are only operated on local units of the image, the geometry and optical deformation of the image are better unchanged, but the overall features of the image are difficult to grasp, and the extraction time is greatly prolonged due to too small feature subdivision.
The SIFT feature extraction method is used for extracting key features of an image by constructing a DOG scale space, searching and positioning key points, direction assignment and key point descriptor generation. Constructing DOG scale space, ensuring that the image has corresponding characteristic points on any scale, and keeping scale invariance; the key point searching and positioning can enhance the noise resistance; the gradient amplitude near the key point has larger weight through Gaussian smoothing in the direction assignment, and the characteristic point instability caused by affine invariance is not considered in the SIFT algorithm; the key point descriptor increases the invariant property of the key points, and improves the efficiency of the matching algorithm. However, the SIFT feature extraction method has fewer feature points for blurred images, and cannot accurately extract features for images with smoother edges, such as fingerprint images. Even though the prior art proposes SURF with better robustness and lower time complexity against the shortcomings of SIFT, the dimension of the feature vector still has 64 dimensions, and the running time cannot meet the needs of practical applications. In addition, the ORB feature extraction method reduces the time complexity by a rapid corner feature detection algorithm, but has poor robustness.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, and provides a fake fingerprint detection system and method based on a composite pixel gradient, which improve the accuracy of judging true and false fingerprints.
In order to achieve the above purpose, the invention is realized by adopting the following technical scheme:
in a first aspect, the present invention provides a fake fingerprint detection method based on a composite pixel gradient, comprising the steps of:
preprocessing a fingerprint image set and extracting an interested region to obtain the interested region of the fingerprint image, and focusing the region of the fingerprint image, in which the fingerprint actually exists, in the fingerprint image in a targeted manner;
dividing a region of interest of a fingerprint image into a plurality of sub-block sets, and calculating the horizontal composite pixel gradient and the vertical composite pixel gradient of adjacent sub-blocks in each sub-block set;
according to the horizontal composite pixel gradient and the vertical composite pixel gradient of adjacent sub-blocks in the sub-block set, calculating the composite pixel gradient amplitude and the composite pixel gradient direction of the sub-block set;
carrying out weighted projection on the composite pixel gradient amplitude of the sub-block set according to the direction of the composite pixel gradient, selecting the direction with the largest weighted projection weight as the main direction of the composite pixel gradient amplitude of the sub-block set, and the secondary direction as the auxiliary direction, so as to ensure that the main direction of the composite pixel gradient amplitude is unchanged no matter how the fingerprint image rotates;
the composite pixel gradient matrix of the sub-block set is obtained through calculation through determining a tuple composed of composite pixel gradient magnitudes in the main direction and the auxiliary direction;
calculating composite pixel gradient matrixes of all sub-block sets and splicing to obtain composite pixel gradient feature matrixes of the interested areas of the fingerprint images;
the method comprises the steps of performing dimension reduction on a plurality of composite pixel gradient feature matrixes, inputting the dimension reduction into a support vector machine, and obtaining a true and false discrimination model of a fingerprint image after training;
and inputting the test fingerprint image into a true and false discrimination model for detection.
With reference to the first aspect, further, the method includes segmenting the region of interest of the fingerprint image into m×n pixel unit blocks, where each k×k pixel unit blocks constitutes a sub-block; the subblocks form a subblock set by a block taking method of self-defining length and width, wherein m and n are positive integers, and k is an odd number.
Further, the method comprises replacing pixel values of the sub-blocks with average values of all pixel unit blocks in one sub-block, weakening interference of noise such as light rays and brightness existing in the fingerprint image, and calculating horizontal composite pixel gradients and vertical composite pixel gradients of adjacent sub-blocks through the replaced pixel values.
Further, the calculation formula of the gradient amplitude of the composite pixel of the sub-block set is as follows:
the direction calculation formula of the composite pixel gradient of the sub-block set is as follows:
θ(x,y)=arctan(G h (x,y)/G v (x,y))
the more complete and comprehensive fingerprint characteristic information is obtained through the calculation formula, wherein M is the gradient amplitude of the composite pixel, theta is the gradient direction of the composite pixel, and G h G is a gradient of horizontal composite pixels v Is a vertical composite pixel gradient.
Further, the method includes performing quantization operations on the horizontal composite pel gradients and the vertical composite pel gradients of all neighboring sub-blocks within a set of sub-blocks, i.e., G v ←G v/Q and Gh ←G h Q, wherein Q is a quantization factor, G h G is a gradient of horizontal composite pixels v For the vertical composite pixel gradient, the composite pixel gradient amplitude dimension reduction of the sub-block set is realized, and the calculated amount is simplified.
Further, the calculation formula of the composite pixel gradient matrix is as follows:
wherein s, t E {0, M 1 },m,n∈{0,M 2 },θ 1 For the main direction of the sub-block set, θ 2 For the auxiliary direction of the set of sub-blocks,is at theta 1 Composite pixel gradient matrix in direction +.>Is at theta 2 Composite pixel gradient matrix in direction, M 1 Is the abscissa of the sub-block set at θ 1 Maximum value taken in direction, M 2 Is the abscissa of the sub-block set at θ 2 Maximum value taken in direction, M θ1 ,M θ2 Respectively, the gradient amplitude M of the composite pixel is within theta 1 ,θ 2 The size of the component in the direction, +.>As a sign function, if x=y, then +.>On the contrary->
In a second aspect, the present invention provides a counterfeit fingerprint detection system based on a gradient of composite pels, said system comprising:
pretreatment and extraction modules: the method comprises the steps of preprocessing a fingerprint image set and extracting an interested region to obtain the interested region of the fingerprint image;
the composite pixel gradient construction module comprises: the method comprises the steps of dividing a region of interest of a fingerprint image into a plurality of sub-block sets, and calculating horizontal composite pixel gradients and vertical composite pixel gradients of adjacent sub-blocks in each sub-block set;
according to the horizontal composite pixel gradient and the vertical composite pixel gradient of adjacent sub-blocks in the sub-block set, calculating the composite pixel gradient amplitude and the composite pixel gradient direction of the sub-block set;
the composite pixel gradient feature matrix construction module comprises: the method comprises the steps of carrying out weighted projection on the gradient amplitude of a composite pixel of a sub-block set according to the direction of the gradient of the composite pixel, and selecting the direction with the largest weight after weighted projection as the main direction of the sub-block set and the secondary direction as the auxiliary direction;
the composite pixel gradient matrix of the sub-block set is obtained through calculation through determining a tuple composed of composite pixel gradient magnitudes in the main direction and the auxiliary direction;
calculating composite pixel gradient matrixes of all sub-block sets and splicing to obtain composite pixel gradient feature matrixes of the interested areas of the fingerprint images;
dimension reduction and training module: the method comprises the steps of performing dimension reduction on a plurality of composite pixel gradient feature matrixes, inputting the dimension reduction into a support vector machine, and obtaining a true and false discrimination model of a fingerprint image after training;
and a detection module: the method is used for inputting the test fingerprint image into the true and false discrimination model for detection.
In a third aspect, the present invention provides a computer device comprising a processor and a storage medium;
the storage medium is used for storing instructions;
the processor is operative to perform the steps of any one of the methods described above in accordance with the instructions.
In a fourth aspect, the present invention provides a computer readable storage medium having stored thereon a computer program for execution by a processor to perform the steps of the method of any one of the preceding claims.
Compared with the prior art, the invention has the following beneficial effects:
the invention carries out pretreatment and region-of-interest extraction on the fingerprint image, extracts the region-of-interest, focuses on the region where the fingerprint actually exists in one fingerprint image, omits other blank parts, and increases the effective region occupation ratio of the composite pixel gradient matrix extracted by subsequent operation;
when the horizontal and vertical composite pixel gradients of adjacent sub-blocks in the sub-block set are calculated, the average value of all pixel unit blocks in the sub-block is used as the pixel value of the sub-block, so that the interference of noise such as light rays, brightness and the like existing in a fingerprint image is effectively weakened;
according to the horizontal composite pixel gradient and the vertical composite pixel gradient of adjacent sub-blocks in the sub-block set, the composite pixel gradient amplitude of the sub-block set is calculated. Compared with the prior art, which only performs feature extraction on the composite pixel gradients of the sub-blocks, the method can obtain more complete and comprehensive fingerprint feature information and improve the classification accuracy of the support vector machine;
the main direction and the auxiliary direction of the gradient amplitude of the composite pixel are determined by utilizing the magnitude of the weighted projection of the gradient amplitude of the composite pixel in the direction of the gradient of the composite pixel, and the main direction of the gradient amplitude of the composite pixel is unchanged no matter how the fingerprint image rotates;
the invention performs dimension reduction operation on the composite pixel gradient feature matrix, realizes the dimension reduction of the feature matrix while keeping enough feature information quantity, and effectively reduces the time for model training of a support vector machine;
the method carries out quantization operation on the horizontal composite pixel gradient and the vertical composite pixel gradient of the sub-block, reduces the dimension of the horizontal composite pixel gradient and the vertical composite pixel gradient of the sub-block through quantization treatment, simultaneously retains as much effective information as possible, further realizes the dimension reduction of the composite pixel gradient amplitude of the sub-block set, and simplifies the calculated amount;
the fake fingerprint detection method based on the composite pixel gradient does not relate to specific image processing experience, is simple in overall method operation, has strong resistance to rotation, illumination and other influence factors, and effectively improves detection efficiency and detection accuracy.
Drawings
FIG. 1 is a flow chart of the detection of fake fingerprints according to an embodiment of the present invention;
fig. 2 is a comparison image of fingerprints before and after ROI extraction provided in an embodiment of the present invention, where a1, a2, a3 are original images, and b1, b2, b3 are respectively extracted images;
FIG. 3 is a conceptual diagram of constructing a composite pixel according to an embodiment of the invention, wherein 3a is the horizontal gradient of a single pixel and 3b is the horizontal gradient of a composite pixel;
fig. 4 is a gradient distribution diagram of a gradient of a composite pixel under different quantization forces, where a-f respectively corresponds to a gradient distribution diagram of a corresponding approximately laplace when 1,2, 4, 8, 16, and 32 are taken as quantization factors Q.
Detailed Description
The present application is described in further detail below with reference to the drawings and examples. The following examples are only for more clearly illustrating the technical solution of the present invention, but should not be construed as limiting the scope of the present invention, and it should be noted that the embodiments and features of the embodiments in the present application may be combined without conflict.
The invention provides a fake fingerprint detection method based on a composite pixel gradient, which comprises the following steps:
preprocessing a fingerprint image set and extracting an interested region to obtain the interested region of the fingerprint image;
dividing a region of interest of a fingerprint image into a plurality of sub-block sets, and calculating the horizontal composite pixel gradient and the vertical composite pixel gradient of adjacent sub-blocks in each sub-block set;
according to the horizontal composite pixel gradient and the vertical composite pixel gradient of adjacent sub-blocks in the sub-block set, calculating the composite pixel gradient amplitude and the composite pixel gradient direction of the sub-block set;
carrying out weighted projection on the composite pixel gradient amplitude of the sub-block set according to the direction of the composite pixel gradient, and selecting the direction with the largest weight after weighted projection as the main direction of the composite pixel gradient amplitude of the sub-block set and the secondary direction as the auxiliary direction;
the composite pixel gradient matrix of the sub-block set is obtained through calculation through determining a tuple composed of composite pixel gradient magnitudes in the main direction and the auxiliary direction;
calculating composite pixel gradient matrixes of all sub-block sets and splicing to obtain composite pixel gradient feature matrixes of the interested areas of the fingerprint images;
the method comprises the steps of performing dimension reduction on a plurality of composite pixel gradient feature matrixes, inputting the dimension reduction into a support vector machine, and obtaining a true and false discrimination model of a fingerprint image after training;
and inputting the test fingerprint image into a true and false discrimination model for detection.
Example 1
As shown in fig. 1, a flow chart of true and false fingerprint detection provided by an embodiment of the present invention is provided, and the detection method includes the following steps: preprocessing and extracting a region of interest from a fingerprint image, constructing a large-scale composite pixel gradient, extracting rotation invariant features, splicing a composite pixel gradient matrix, reducing the dimension of a feature matrix PCA and training a classifier model.
And inputting the test fingerprint image into a trained classifier model, and judging whether the test fingerprint image is true or false by the classifier model.
The detailed flow of the above steps is specifically described below:
preprocessing and extracting a region of interest: converting the numerical values of three components of the fingerprint image R, G, B in the training set according to weights by a weighted average method to obtain a fingerprint gray image; and (3) performing edge detection on the fingerprint gray level image by using a Sobel operator, solving a minimum edge circumscribed rectangle, and intercepting an interested region of the fingerprint image.
The weighted average method is calculated as follows:
Gray(i,j)=0.299*R(i,j)+0.578*G(i,j)+0.114*B(i,j)
wherein: gray (i, j) represents the image Gray value of the pixel after weighted averaging, and R (i, j), G (i, j), B (i, j) respectively represent the sizes of three components of the fingerprint image R, G, B at the pixel.
The algorithm for edge detection using the Sobel operator is as follows:
wherein: gx represents the image gray value detected by the lateral edge, gy represents the image gray value detected by the longitudinal edge, G represents the gradient magnitude, Θ represents the gradient direction, and a represents the pixel value matrix of the original image.
Fig. 2 shows fingerprint comparison images before and after ROI extraction provided in the embodiment of the present invention, where a1, a2, a3 are original images, b1, b2, b3 are extracted images respectively, and as can be seen from fig. 2, weighted average graying processing and region of interest (ROI) extraction, the fingerprint images are converted into gray images and are collectively represented as required fingerprint portions, the white regions are reduced, interference of the white regions on the classifier finally trained in the feature learning process is reduced, and accuracy of classifying the subsequent classifier is improved.
Constructing a large-scale composite pixel gradient: the pixel gradient can only represent the pixel value information in a unit pixel neighborhood range, such as a single pixel shown in fig. 3a, the pixel gradient is greatly influenced by external noise, and the pixel value information in a unit pixel neighborhood range (window size is 3×3) cannot represent a relatively complete fingerprint structure. Wherein 3a is the horizontal gradient of a single pixel and 3b is the horizontal gradient of a composite pixel.
As shown in fig. 3, the present invention first divides the region of interest F in the fingerprint image into m×n pixel unit blocks, each k×k pixel unit blocks form a sub-block, and the sub-blocks form a sub-block set by a user-defined length and width block taking method. Wherein m, n are positive integers and k is an odd number.
In the embodiment of the present invention, k is 3, and the self-defined aspect ratio is 3×3, that is, 3a is a sub-block set in this embodiment.
As shown in fig. 3a, the average value of each sub-block in the set of sub-blocks is calculated and denoted μ;
finally, the horizontal composite pixel gradient G of the adjacent subblocks in the subblock set is counted h And vertical composite pixel gradient G v The calculation formula is as follows:
in the formula, "1" represents one sub-block.
And similarly, calculating the horizontal composite pixel gradient and the vertical composite pixel gradient of all adjacent sub-blocks in the sub-block set.
From this, the composite pixel gradient amplitude M and direction θ of the sub-block set can be calculated, and the calculation formulas are respectively:
θ(x,y)=arctan( h (x,y)/G v (x,y))
wherein M is the gradient amplitude of the composite pixel, θ is the gradient direction of the composite pixel, G h G is a gradient of horizontal composite pixels v Is a vertical composite pixel gradient.
In this embodiment, the gradient G of the horizontal composite pixel h And vertical composite pixel gradient G v The value ranges are [ -255,255]。
In order to reduce the complex calculation amount caused by overlarge feature dimension, after further researching the influence of different quantization forces on the composite pixel gradients of the sub-blocks on the composite pixel gradient amplitude M of the sub-block set, the invention executes quantization operation on the composite pixel gradients of all the sub-blocks, namely G v ←G v Q or G h ←G h Q, Q is quantization factor, G h G is a gradient of horizontal composite pixels v Is a vertical composite pixel gradient.
Fig. 4 is a graph showing gradient profiles of approximately laplace obtained under different quantization factors Q according to an embodiment of the present invention, and a-f respectively correspond to the gradient profiles of approximately laplace when 1,2, 4, 8, 16, and 32 are taken for the quantization factors Q. As can be seen from fig. 4: along with the increase of the quantization factor Q, the gradient value range of the composite pixel gradually becomes smaller, so that the dimension reduction of the gradient of the composite pixel can be realized while retaining as much effective information as possible, the dimension reduction of the gradient amplitude M of the composite pixel of the sub-block set is realized, and the calculated amount is simplified.
Extracting rotation invariant features and splicing a composite pixel gradient matrix: and calculating the gradient amplitude and the gradient direction of the composite pixel of all sub-block sets in the region of interest of the fingerprint image according to the steps, wherein the range of the gradient and the direction is recorded as (0, pi). The operation of one sub-block set is specifically described below, and the same applies.
Firstly, equally dividing the direction theta of the sub-block set into four parts; then, the methodThe gradient amplitude M of the composite pixel of the sub-block set is subjected to weighted projection in four directions equally divided by theta; finally, selecting the gradient direction theta with the largest weight 1 For the main direction of the sub-block set, the next-largest direction θ 2 As an auxiliary direction.
The composite pixel gradient feature matrix of the sub-block set is represented by a tuple consisting of the following two composite pixel gradient amplitudes and directions, and the calculation formula is as follows:
wherein s, t E {0, M 1 },m,n∈{0,M 2 },θ 1 For the main direction of the sub-block set, θ 2 For the auxiliary direction of the set of sub-blocks,is at theta 1 Composite pixel gradient matrix in direction +.>Is at theta 2 Composite pixel gradient matrix in direction, M 1 Is the abscissa of the sub-block set at θ 1 Maximum value taken in direction, M 2 Is the abscissa of the sub-block set at θ 2 Maximum value taken in direction, M θ1 ,M θ2 Respectively, the gradient amplitude M of the composite pixel is within theta 1 、θ 2 The size of the component in the direction, +.>As a sign function, if x=y, then +.>On the contrary->
In order to prevent the classifier from carrying out a large amount of inner product operation and simultaneously reduce the interference of abnormal values in the composite pixel gradient feature matrix, the invention executes normalization processing on the composite pixel gradient feature matrix and restricts the range of each element value in the composite pixel gradient feature matrix to (0, 1).
By adopting the same operation, the composite pixel gradient feature matrix of all sub-block sets in the fingerprint image can be obtained. And finally, splicing the composite pixel gradient feature matrixes of different sub-block sets to obtain the composite pixel gradient feature matrix of the fingerprint image, and taking the composite pixel gradient feature matrix as the input of the classifier.
Feature matrix PCA dimension reduction: the dimension reduction mainly utilizes a principal component analysis technology, wherein the principal component analysis technology mainly reduces the dimension of a composite pixel gradient feature matrix of a fingerprint image, and simultaneously reserves the feature with the largest contribution to the whole variance in the composite pixel gradient feature matrix, and the calculation method comprises the following steps:
wherein i represents the row label of the composite pixel gradient feature matrix, j represents the column label of the composite pixel gradient feature matrix, and x ij Is an element with the position of i row and j column in the gradient feature matrix of the composite pixel, r is a correlation matrix, n is the number of original features, Z is a matrix formed by dividing the difference value of each column by the standard deviation of the column, Z T Is the transposed matrix of Z.
After the correlation matrix R is obtained, the characteristic equation |R-lambda I of the correlation matrix R is solved p I=0, p feature roots are obtained, λ is the weight, I p Is the characteristic root. Pressing the buttonDetermining m value, wherein m is the final main component feature quantity, so that the information utilization rate reaches more than 85% in the m main component features; for a pair ofEach lambda is j Solving equation set R b =λj b Obtaining unit feature vector->Where j=1, 2,..m, λ is the weight.
The variables obtained above are put into the following principal component feature matrix calculation formula:
wherein Uij To obtain a principal component feature matrix.
The features extracted by the embodiment of the invention are subjected to principal component analysis, the feature dimension is reduced to two dimensions or three dimensions, and the obtained features are obviously divided.
Training a classifier model: the invention selects a classifier using a support vector machine as a characteristic, and the optimization method of the support vector machine linear separable problem is as follows:
in the formula ,ai ,a j Are Lagrangian, x i ,x j Divided into i and j feature vectors,is x i Transposed matrix of y i ,y j All are class labels, and n is the number of rows of the principal component feature matrix obtained above.
Screening the super parameters by adopting a grid search method, specifically performing enumeration search according to different appointed super parameter lists, evaluating the accuracy and stability of the classifier model under each group of super parameter combinations, and selecting the parameter combination with the highest accuracy and better stability as the super parameter of the classifier model, wherein the classifier model is used for judging whether the fingerprint to be detected is true or false.
The detection results of the present invention are described below for a common data set. The fingerprint image library used in the invention is from LivDet2017 of international fingerprint activity detection (artificial fingerprint detection) competition. The data set is formed by combining images acquired by three scanners, and the images of an image library are divided into two parts which are not overlapped: the training set and the testing set are used for training and testing the classifier respectively.
The material for manufacturing the fake fingerprint training set comprises wood glue, resin and other materials, and the material for manufacturing the fake fingerprint test set comprises gelatin, latex and other materials. The basic situation of the LivDet2017 fingerprint image is shown in table 1.
Table 1: livDet2017 fingerprint image database basic information
In the experiment of the algorithm, the sklearn framework is used, the packaging degree is higher, and the method has a perfect library for realizing the classification problem, so that the experiment classification is facilitated.
The invention utilizes a feature extraction algorithm based on composite pixel gradients and SVM kernel classification training, and the obtained final test results are shown in Table 2.
Table 2: final test results
As can be seen from table 2: the accuracy of the classifier model trained by the invention on the test set is improved by about 3 percent compared with the accuracy of the test result of extracting the characteristics by using a convolutional neural network in the prior art; meanwhile, the variance is smaller in the training set, and the classifier model is more stable.
Example two
A counterfeit fingerprint detection system based on a composite pel gradient, the system comprising:
pretreatment and extraction modules: the method comprises the steps of preprocessing a fingerprint image set and extracting an interested region to obtain the interested region of the fingerprint image;
the composite pixel gradient construction module comprises: the method comprises the steps of dividing a region of interest of a fingerprint image into a plurality of sub-block sets, and calculating horizontal composite pixel gradients and vertical composite pixel gradients of adjacent sub-blocks in each sub-block set;
according to the horizontal composite pixel gradient and the vertical composite pixel gradient of adjacent sub-blocks in the sub-block set, calculating the composite pixel gradient amplitude and the composite pixel gradient direction of the sub-block set;
the composite pixel gradient feature matrix construction module comprises: the method comprises the steps of carrying out weighted projection on the gradient amplitude of a composite pixel of a sub-block set according to the direction of the gradient of the composite pixel, and selecting the direction with the largest weight after weighted projection as the main direction of the sub-block set and the secondary direction as the auxiliary direction;
the composite pixel gradient matrix of the sub-block set is obtained through calculation through determining a tuple composed of composite pixel gradient magnitudes in the main direction and the auxiliary direction;
calculating composite pixel gradient matrixes of all sub-block sets and splicing to obtain composite pixel gradient feature matrixes of the interested areas of the fingerprint images;
dimension reduction and training module: the method comprises the steps of performing dimension reduction on a plurality of composite pixel gradient feature matrixes, inputting the dimension reduction into a support vector machine, and obtaining a true and false discrimination model of a fingerprint image after training;
and a detection module: the method is used for inputting the test fingerprint image into the true and false discrimination model for detection.
Example III
The invention provides a computer device, comprising a processor and a storage medium;
the storage medium is used for storing instructions;
the processor is operative in accordance with the instructions to perform the steps of a counterfeit fingerprint detection method based on a gradient of composite pels.
Example IV
The invention also provides a computer readable storage medium having stored thereon a computer program for execution by a processor to perform the steps of a method of detecting a counterfeit fingerprint based on a gradient of composite pels.
The invention provides a fake fingerprint detection system and a fake fingerprint detection method based on composite pixel gradients, which are characterized in that firstly, preprocessing is carried out on a fingerprint image and characteristics of an interested region are extracted, then, a model framework for extracting the characteristics of the fingerprint is provided from two aspects of construction of the composite pixel gradients and rotation-unchanged characteristic statistics, detection of true and false fingerprints is realized, the method is simple and easy to implement, and compared with the prior art, the accuracy of judging the true and false fingerprints is improved, and the method has obvious technical progress and market application value.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, systems, and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The scope of the invention is obviously not limited to these specific embodiments. Equivalent modifications and substitutions for related technical features may be made by those skilled in the art without departing from the principles of the present invention, and such modifications and substitutions will fall within the scope of the present invention.
Claims (9)
1. A fake fingerprint detection method based on composite pixel gradient is characterized by comprising the following steps:
preprocessing a fingerprint image set and extracting an interested region to obtain the interested region of the fingerprint image;
dividing a region of interest of a fingerprint image into a plurality of sub-block sets, and calculating the horizontal composite pixel gradient and the vertical composite pixel gradient of adjacent sub-blocks in each sub-block set;
according to the horizontal composite pixel gradient and the vertical composite pixel gradient of adjacent sub-blocks in the sub-block set, calculating the composite pixel gradient amplitude and the composite pixel gradient direction of the sub-block set;
carrying out weighted projection on the composite pixel gradient amplitude of the sub-block set according to the direction of the composite pixel gradient, and selecting the direction with the largest weight after weighted projection as the main direction of the composite pixel gradient amplitude of the sub-block set and the secondary direction as the auxiliary direction;
the composite pixel gradient matrix of the sub-block set is obtained through calculation through determining a tuple composed of composite pixel gradient magnitudes in the main direction and the auxiliary direction;
calculating composite pixel gradient matrixes of all sub-block sets and splicing to obtain composite pixel gradient feature matrixes of the interested areas of the fingerprint images;
the method comprises the steps of performing dimension reduction on a plurality of composite pixel gradient feature matrixes, inputting the dimension reduction into a support vector machine, and obtaining a true and false discrimination model of a fingerprint image after training;
and inputting the test fingerprint image into a true and false discrimination model for detection.
2. A counterfeit fingerprint detection method according to claim 1, wherein said method comprises segmenting a region of interest of a fingerprint image intoA pixel unit block, each->The pixel unit blocks form a sub-block; the subblocks form a subblock set by a block taking method of self-defining length and width, wherein +_>, />Is a positive integer, and k is an odd number.
3. A fake fingerprint detection method according to claim 2, characterized in that the method comprises replacing the pixel value of a sub-block with an average value of all pixel unit blocks within the sub-block, and calculating the horizontal and vertical composite pixel gradients of adjacent sub-blocks from the replaced pixel values.
4. The method for detecting counterfeit fingerprints according to claim 1, wherein the calculation formula of the gradient magnitude of the composite pixel of the sub-block set is:
the direction calculation formula of the composite pixel gradient of the sub-block set is as follows:
5. A counterfeit fingerprint detection method according to claim 1, wherein said method comprises performing quantization operations on horizontal and vertical composite pel gradients of all adjacent sub-blocks within a set of sub-blocks, namely and />, wherein />As a quantization factor, < >>Is a horizontal composite pixel gradient->Is a vertical composite pixel gradient.
6. The method for detecting counterfeit fingerprints according to claim 1, wherein the composite pixel gradient matrix has a calculation formula:
wherein ,,/>,/>for the main direction of the set of sub-blocks +.>For the auxiliary direction of the set of sub-blocks, +.>Is at->Composite pixel gradient matrix in direction +.>Is atComposite pixel gradient matrix in direction +.>Is the abscissa of the sub-block set +.>The maximum value that is achieved in the direction,is the abscissa of the sub-block set +.>Maximum value taken in direction, +_>Respectively the gradient amplitude of the composite pixelAt->,/>The size of the component in the direction, +.>As a sign function if->Then->The method comprises the steps of carrying out a first treatment on the surface of the Otherwise。
7. A counterfeit fingerprint detection system based on a gradient of composite pels, said system comprising:
pretreatment and extraction modules: the method comprises the steps of preprocessing a fingerprint image set and extracting an interested region to obtain the interested region of the fingerprint image;
the composite pixel gradient construction module comprises: the method comprises the steps of dividing a region of interest of a fingerprint image into a plurality of sub-block sets, and calculating horizontal composite pixel gradients and vertical composite pixel gradients of adjacent sub-blocks in each sub-block set;
according to the horizontal composite pixel gradient and the vertical composite pixel gradient of adjacent sub-blocks in the sub-block set, calculating the composite pixel gradient amplitude and the composite pixel gradient direction of the sub-block set;
the composite pixel gradient feature matrix construction module comprises: the method comprises the steps of carrying out weighted projection on the gradient amplitude of a composite pixel of a sub-block set according to the direction of the gradient of the composite pixel, and selecting the direction with the largest weight after weighted projection as the main direction of the sub-block set and the secondary direction as the auxiliary direction;
the composite pixel gradient matrix of the sub-block set is obtained through calculation through determining a tuple composed of composite pixel gradient magnitudes in the main direction and the auxiliary direction;
calculating composite pixel gradient matrixes of all sub-block sets and splicing to obtain composite pixel gradient feature matrixes of the interested areas of the fingerprint images;
dimension reduction and training module: the method comprises the steps of performing dimension reduction on a plurality of composite pixel gradient feature matrixes, inputting the dimension reduction into a support vector machine, and obtaining a true and false discrimination model of a fingerprint image after training;
and a detection module: the method is used for inputting the test fingerprint image into the true and false discrimination model for detection.
8. A computer device comprising a processor and a storage medium;
the storage medium is used for storing instructions;
the processor being operative according to the instructions to perform the steps of the method according to any one of claims 1-6.
9. A computer readable storage medium having stored thereon a computer program, characterized in that the computer program is executed by a processor to perform the steps of the method according to any of claims 1-6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310116481.1A CN115841685B (en) | 2023-02-15 | 2023-02-15 | Fake fingerprint detection system and method based on composite pixel gradient |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310116481.1A CN115841685B (en) | 2023-02-15 | 2023-02-15 | Fake fingerprint detection system and method based on composite pixel gradient |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115841685A CN115841685A (en) | 2023-03-24 |
CN115841685B true CN115841685B (en) | 2023-05-12 |
Family
ID=85579682
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310116481.1A Active CN115841685B (en) | 2023-02-15 | 2023-02-15 | Fake fingerprint detection system and method based on composite pixel gradient |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115841685B (en) |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104268529A (en) * | 2014-09-28 | 2015-01-07 | 深圳市汇顶科技股份有限公司 | Judgment method and device for quality of fingerprint images |
CN108520225B (en) * | 2018-03-30 | 2021-07-27 | 南京信息工程大学 | Fingerprint detection classification method based on spatial transformation convolutional neural network |
CN109063572A (en) * | 2018-07-04 | 2018-12-21 | 南京信息工程大学 | It is a kind of based on multiple dimensioned and multireel lamination Fusion Features fingerprint activity test methods |
CN109255318A (en) * | 2018-08-31 | 2019-01-22 | 南京信息工程大学 | Based on multiple dimensioned and multireel lamination Fusion Features fingerprint activity test methods |
-
2023
- 2023-02-15 CN CN202310116481.1A patent/CN115841685B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN115841685A (en) | 2023-03-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Masci et al. | Steel defect classification with max-pooling convolutional neural networks | |
Satpathy et al. | LBP-based edge-texture features for object recognition | |
CN105447441B (en) | Face authentication method and device | |
CN112686812B (en) | Bank card inclination correction detection method and device, readable storage medium and terminal | |
CN105825183B (en) | Facial expression recognizing method based on partial occlusion image | |
CN105678788B (en) | A kind of fabric defect detection method based on HOG and low-rank decomposition | |
Jain et al. | Learning to agglomerate superpixel hierarchies | |
Grundmann et al. | 3D shape context and distance transform for action recognition | |
Zhou et al. | Histograms of categorized shapes for 3D ear detection | |
CN104217221A (en) | Method for detecting calligraphy and paintings based on textural features | |
CN105701495B (en) | Image texture feature extraction method | |
CN104036244B (en) | Checkerboard pattern corner point detecting method and device applicable to low-quality images | |
CN112488211A (en) | Fabric image flaw classification method | |
Yang et al. | Log-euclidean kernel-based joint sparse representation for hyperspectral image classification | |
Avola et al. | Real-time deep learning method for automated detection and localization of structural defects in manufactured products | |
CN106874942A (en) | A kind of object module fast construction method semantic based on regular expression | |
Velliangira et al. | A novel forgery detection in image frames of the videos using enhanced convolutional neural network in face images | |
Qi et al. | Exploring illumination robust descriptors for human epithelial type 2 cell classification | |
Ali et al. | Speeded up robust features for efficient iris recognition | |
Liao et al. | Unconstrained face detection | |
CN111127407B (en) | Fourier transform-based style migration forged image detection device and method | |
Diaa | A Deep Learning Model to Inspect Image Forgery on SURF Keypoints of SLIC Segmented Regions | |
CN115841685B (en) | Fake fingerprint detection system and method based on composite pixel gradient | |
Li et al. | Real and fake label image classification algorithm based on hog and svm | |
Makandar et al. | Transform Domain Techniques combined with LBP for Tampered Image Detection using Machine Learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |