CN111597895A - OCT fingerprint anti-counterfeiting method based on resnet50 - Google Patents
OCT fingerprint anti-counterfeiting method based on resnet50 Download PDFInfo
- Publication number
- CN111597895A CN111597895A CN202010293470.7A CN202010293470A CN111597895A CN 111597895 A CN111597895 A CN 111597895A CN 202010293470 A CN202010293470 A CN 202010293470A CN 111597895 A CN111597895 A CN 111597895A
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- oct
- resnet50
- local small
- finger
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000012014 optical coherence tomography Methods 0.000 claims abstract description 45
- 238000012549 training Methods 0.000 claims abstract description 34
- 238000013528 artificial neural network Methods 0.000 claims abstract description 9
- 238000007781 pre-processing Methods 0.000 claims abstract description 9
- 210000000434 stratum corneum Anatomy 0.000 claims abstract description 8
- 238000012360 testing method Methods 0.000 claims abstract description 8
- 230000006870 function Effects 0.000 claims description 20
- 239000000463 material Substances 0.000 claims description 16
- 239000004831 Hot glue Substances 0.000 claims description 15
- 239000000853 adhesive Substances 0.000 claims description 6
- 230000001070 adhesive effect Effects 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 6
- 210000000106 sweat gland Anatomy 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 238000012935 Averaging Methods 0.000 claims description 3
- 108010010803 Gelatin Proteins 0.000 claims description 3
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N Silicium dioxide Chemical compound O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 claims description 3
- 238000002474 experimental method Methods 0.000 claims description 3
- 239000000499 gel Substances 0.000 claims description 3
- 239000008273 gelatin Substances 0.000 claims description 3
- 229920000159 gelatin Polymers 0.000 claims description 3
- 235000019322 gelatine Nutrition 0.000 claims description 3
- 235000011852 gelatine desserts Nutrition 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000002844 melting Methods 0.000 claims description 3
- 230000008018 melting Effects 0.000 claims description 3
- 238000011176 pooling Methods 0.000 claims description 3
- 238000003825 pressing Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000007493 shaping process Methods 0.000 claims description 3
- 239000000741 silica gel Substances 0.000 claims description 3
- 229910002027 silica gel Inorganic materials 0.000 claims description 3
- 239000003990 capacitor Substances 0.000 claims description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 210000003491 skin Anatomy 0.000 description 2
- 238000013475 authorization Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000003292 glue Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 239000010985 leather Substances 0.000 description 1
- 210000001595 mastoid Anatomy 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1382—Detecting the live character of the finger, i.e. distinguishing from a fake or cadaver finger
- G06V40/1388—Detecting the live character of the finger, i.e. distinguishing from a fake or cadaver finger using image processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1347—Preprocessing; Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1365—Matching; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Collating Specific Patterns (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
An OCT fingerprint anti-counterfeiting method based on a resnet50 neural network comprises the following steps: the first step is as follows: preparing various types of fingerprint samples, and collecting OCT (optical coherence tomography) volume data corresponding to the fingerprint samples; the second step is that: preprocessing the acquired data; the third step: performing first-order longitudinal differential operation on a section image obtained from the acquired OCT volume data, obtaining the position of each row of characteristic points, obtaining a stratum corneum after denoising and fitting of a connected domain, selecting a certain characteristic point on the stratum corneum position, and taking a local small block as a network training sample; the fourth step: constructing a resnet50 network model; the fifth step: in the process of identifying OCT volume data randomly, local small blocks are input into a trained resnet50 network model, the category of the local small blocks is determined, and anti-counterfeiting is carried out. The invention carries out data classification on the test data through the deep-level information of the finger learned by the network, thereby achieving fingerprint anti-counterfeiting.
Description
Technical Field
The invention relates to the field of fingerprint anti-counterfeiting, in particular to an OCT fingerprint anti-counterfeiting method based on a resnet50 neural network.
Background
The fingerprint anti-counterfeiting technology specifically refers to detecting demonstration attacks of fingerprints and judging the properties of the demonstration attacks. The demonstration attacks of fingerprints are mainly represented by fingerprint films, full-finger counterfeits, broken fingers, masks and some system fraud authorizations. The unique image characteristics of the image trace of the fingerprint become the basic conditions for fingerprint counterfeiting. Meanwhile, the mastoid patterns on the surface of the finger have stable shape and volume, so that the referential and moldable original conditions for fingerprint counterfeiting are achieved. Most of the existing popular fingerprint counterfeiting is based on soft rubber materials with similar properties to the skin of the finger, and the simulated fake sample which has the same lines as the real finger and also has similar optical, electrical and mechanical properties is obtained. This can easily break through most of the fingerprint readers on the market.
Optical Coherence Tomography (OCT), a non-invasive imaging technique, scans 2-3mm of information under the skin of a finger. Meanwhile, the papillary layer and sweat gland tissues of the finger are positioned in the range, and the papillary layer of the finger is verified to be a genuine leather fingerprint source of the finger and has the same fingerprint lines as the external fingerprint lines of the finger. Development of OCT technology and results brought by the development open a new door of fingerprint anti-counterfeiting in the field of fingerprint identification.
Disclosure of Invention
In order to overcome the demonstration attack brought by the existing fingerprint counterfeit sample, the invention provides an OCT fingerprint anti-counterfeiting method based on the resnet50 neural network, after preprocessing a slice image (B-scan image) acquired by an OCT system, taking a local small block containing effective information, inputting the local small block into the network for training, and the obtained training result can rapidly and automatically realize the classification of fingerprints, thereby realizing the fingerprint anti-counterfeiting function.
In order to achieve the purpose, the invention adopts the technical scheme that:
an OCT fingerprint anti-counterfeiting method based on resnet50 comprises the following steps:
1) the soft rubber material with similar properties to the fingers is utilized: the method comprises the following steps of (1) imitating fingerprint films and fake finger samples with finger fingerprints by using silica gel, gelatin and capacitor gel, and acquiring OCT (optical coherence tomography) volume data of real fingers and fake fingerprint samples by using an OCT (optical coherence tomography) system;
2) filtering and preprocessing data of the slices of different fingerprint samples to obtain smooth data;
3) performing first-order longitudinal differential operation on the slice image, obtaining the position of each row of feature points, obtaining the stratum corneum after denoising and fitting the connected domain, determining the feature points on the stratum corneum, and taking local small blocks with the size of 224 x 224 according to 30 upward pixel points and 112 left and right pixel points of the feature points;
4) training and dividing a test set according to the proportion of 8:2 of a data set, constructing a resnet50 network model, setting training parameters and a loss function, and training a resnet50 network model by using the obtained local small blocks with the size of 224 × 224 to obtain a trained resnet50 network model;
5) and (3) randomly selecting feature points in 4B-scan graphs of the data to be identified to obtain local small blocks with the size of 224 x 224, inputting the obtained local small blocks into a trained resnet50 network model, determining the category of the 4 samples according to probability classification of the 4 samples, and performing anti-counterfeiting on the samples.
Further, the step 1) comprises the following steps:
(11) melting the hot melt adhesive, injecting the hot melt adhesive into an experiment platform, slightly pressing a tested finger on the hot melt adhesive when the temperature of the hot melt adhesive is reduced to 42 ℃ and the hot melt adhesive is not solidified, and obtaining a female die;
(12) preparing the prepared soft adhesive material, and injecting the soft adhesive material into the female die to obtain a fingerprint film with the thickness of 0.1-1 mm;
(13) placing the female die in a shaping material similar to a finger, injecting a soft rubber material, and solidifying to obtain an artificial finger sample with a fingerprint;
(14) the obtained fingerprint samples are acquired 3 times in 2 different periods respectively by using an OCT system, and OCT volume data of the fingerprint samples are obtained.
Still further, the step 2) comprises the following steps:
(21) filtering the B-scan image acquired by the OCT by using BM3D to remove noise points in the acquisition process;
(22) the filtered image is subjected to the following preprocessing operation.
Where img _ temp is the pixel value of the image, img _ mean is the mean of the image pixels, and img _ std is the standard deviation of the image.
Further, the step 3) comprises the following steps:
(31) by a first longitudinal difference Iy(x, y) to approximate the feature points of the extracted image I:
Iy(x,y)＝I(x+1,y)-I(x,y) (2)
where (x, y) is the pixel coordinate, and I (x, y) represents the gray level of the pixel located at the coordinate (x, y), then each column I is selectedyThe point with the maximum (x, y) value is taken as a characteristic point;
(32) carrying out connected domain denoising on feature points of a 500 x 1800 section image to obtain a position of a finger horny layer, selecting a certain feature point on the position of the horny layer, taking 30 pixels upwards according to the feature points, taking 112 pixels from the left and right, and taking a local small block with the size of 224 x 224, so that the local small block taken by training can be ensured to contain horny layer, papillary layer and sweat gland tissue information, sequentially spacing 50 pixels, and taking the local small block to obtain a sufficient training sample;
(33) and (5) repeating the steps (31) and (32), and carrying out OCT slice images on the real finger, the fingerprint film sample and the full finger sample to carry out local small block sampling.
Further, the step 4) comprises the following steps:
(41) taking 80% of the data obtained in the step 3 as a training set and 20% as a test set, wherein the fingerprint images of the real fingers are stored in an img _1 folder, the fingerprint film images are stored in an img _2 folder, and the fake finger images are stored in an img _3 folder;
(42) inputting training samples into a Resnet50 neural network, wherein the Resnet50 network has 50 convolutional layers, 50 ReLU layers, 2 pooling layers, 1 fully-connected layer, and the main structure of Resnet50 is a residual structure, h (x) ═ f (x) + x, if the output of the shallow network is mature enough, i.e. the learning f (x) is equal to 0, this time, it is an identity mapping;
(43) fully connecting the residual module with the features extracted from each convolution layer of the network, specifically classifying the fingerprints by utilizing softmax, using an Adam optimizer as an optimizer, using a cross entropy Loss function as a Loss, loading 4 pictures in a training set into a resnet50 neural network for training, and obtaining the trained network with the iteration number of 100 times;
in the step (43), a cross entropy loss function based on softmax is used; the softmax function is of the form:
in the above formula, ak(x) Representing the output value of the kth class at point x,
representing the sum of the output values of all classes at point x, Pk(x) Represents the probability of class k at point x;
the cross entropy loss function is of the form:
where w (x) represents the weight parameter of the model, l (x) represents the true class of point x, Pl(x)(x) Representing the probability of the true class at point x and E representing the loss value of the cross entropy function.
Further, the step 5) comprises the following steps:
(51) randomly selecting 4B-scan images X1, X2, X3 and X4 from the OCT volume data to be identified, and randomly selecting one local small block P1, P2, P3 and P4 from the 4B-scan images according to the step 3);
(52) inputting 4 local small blocks P1, P2, P3 and P4 into a trained network, classifying the probability of the 4 local small blocks, averaging P to (P1+ P2+ P3+ P4)/4, and realizing the category to which the final classification belongs, thereby achieving the purpose of fingerprint anti-counterfeiting.
The invention has the beneficial effects that: after preprocessing a slice image (B-scan image) acquired by an OCT system, taking a local small block containing effective information, inputting the local small block into a network for training, and quickly and automatically realizing the classification of fingerprints according to an obtained training result so as to realize the fingerprint anti-counterfeiting function.
Drawings
FIG. 1 is a flow chart of fingerprint authentication according to the present invention;
FIG. 2 is a schematic diagram of local tile extraction in the present invention;
Detailed Description
The invention will be further described with reference to the following figures and embodiments:
referring to fig. 1 and 2, an OCT fingerprint anti-counterfeiting method based on resnet50 includes the following steps:
1) probing the properties of the finger: including refractive index, hardness, conductivity, to obtain a soft gel material with similar properties: silica gel, gelatin, electric capacity glue imitate fingerprint film and the artificial finger sample that has the finger fingerprint, utilize OCT system, gather all fingerprint sample's OCT volume data, include the following step:
(11) melting the hot melt adhesive, injecting the hot melt adhesive into an experiment platform, slightly pressing a tested finger on the hot melt adhesive when the temperature of the hot melt adhesive is reduced to be below 42 ℃ and the hot melt adhesive is not solidified, and obtaining a female die;
(12) preparing the prepared soft adhesive material, and injecting the soft adhesive material into the female die to obtain a fingerprint film with the thickness of 0.1-1 mm;
(13) placing the female die in a shaping material containing fingers, injecting a soft rubber material, and solidifying to obtain an artificial finger sample with fingerprints;
(14) acquiring the obtained fingerprint sample 3 times in 2 different time periods respectively by using an OCT system to obtain OCT volume data;
2) the method comprises the following steps of filtering and preprocessing data of slices of different fingerprint samples acquired to obtain smooth data, and accelerating the speed of solving an optimal solution by gradient descent, and comprises the following steps:
(21) filtering the B-scan image acquired by the OCT by using BM3D to remove noise points in the acquisition process;
(22) preprocessing the filtered image, accelerating the speed of solving an optimal solution by gradient descent, and adding 1 to the values of all pixel points of the image and amplifying by 25 times to enable the image information to be more obvious in order to avoid the occurrence of pixel points smaller than 0 in the image due to the fact that the standard normal distribution data is symmetrical about x to 0;
wherein img _ temp is a pixel value of an image, img _ mean is a mean value of pixels of the image, and img _ std is a standard deviation of the image;
3) the method comprises the following steps of performing first-order longitudinal differential operation on a section picture, obtaining the position of each row of feature points, obtaining the stratum corneum after denoising and fitting of a connected domain, upwards moving 30 pixel points according to the feature point positions on the stratum corneum, controlling 112 pixel points, and obtaining local small blocks with the size of 224 x 224, wherein the method comprises the following steps:
(31) by a first longitudinal difference Iy(x, y) to approximate the feature points of the extracted image I:
Iy(x,y)＝I(x+1,y)-I(x,y) (2)
where (x, y) is the pixel coordinate, and I (x, y) represents the gray value of the pixel located at the coordinate (x, y). Then selecting each column IyThe point with the maximum (x, y) value is taken as a characteristic point;
(32) carrying out connected domain denoising on feature points of a 500 x 1800 section image to obtain a position of a finger horny layer, selecting a certain feature point on the position of the horny layer, taking 30 pixels upwards according to the feature points, taking 112 pixels from the left and right, and taking a local small block with the size of 224 x 224, so that the local small block taken by training can be ensured to contain horny layer, papillary layer and sweat gland tissue information, sequentially spacing 50 pixels, and taking the local small block to obtain a sufficient training sample;
(33) repeating the steps (31) and (32), and carrying out local small-block sampling on the OCT slice images of the objects such as the real finger, the fingerprint film sample, the full finger sample and the like so as to ensure that enough characteristics and enough training samples exist;
4) training and dividing a test set according to an 8:2 ratio, constructing a resnet50 network model, setting training parameters and a loss function, training a resnet50 network model by using the obtained local small blocks with the size of 224 × 224, and obtaining a trained resnet50 network model, wherein the method comprises the following steps:
(41) taking 80% of the data obtained in the step 3 as a training set and 20% as a test set, wherein the fingerprint images of the real fingers are stored in an img _1 folder, the fingerprint film images are stored in an img _2 folder, and the fake finger images are stored in an img _3 folder;
(42) inputting training samples into a Resnet50 neural network, wherein the Resnet50 network has 50 convolutional layers, 50 ReLU layers, 2 pooling layers, 1 fully-connected layer, and the main structure of Resnet50 is a residual structure, h (x) ═ f (x) + x, if the output of the shallow network is mature enough, i.e. the learning f (x) is equal to 0, this time, it is an identity mapping;
(43) fully connecting the residual module with the features extracted from each convolution layer of the network, specifically classifying the fingerprints by utilizing softmax, using an Adam optimizer as an optimizer, using a cross entropy Loss function as a Loss, loading 4 pictures in a training set into a resnet50 neural network for training, and obtaining the trained network with the iteration number of 100 times;
in the step (43), a cross entropy loss function based on softmax is used; the softmax function is of the form:
in the above formula, ak(x) Representing the output value of the kth class at point x,
representing the sum of the output values of all classes at point x, Pk(x) Represents the probability of class k at point x;
the cross entropy loss function is of the form:
where w (x) represents the weight parameter of the model, l (x) represents the true class of point x, pl(x)(x) Representing the probability of the real category of the point x, and E represents the loss value of the cross entropy function;
5) inputting a prepared test picture, randomly selecting 4B-scan pictures to obtain local small blocks with the size of 224 x 224, inputting the local small blocks into a trained resnet50 network model, determining the category of the local small blocks according to probability classification of 4 samples, and performing anti-counterfeiting on the local small blocks, wherein the method comprises the following steps:
(51) randomly selecting 4B-scan images in the OCT volume data, and randomly taking one local small block in the 4B-scan images according to the step 3);
(52) inputting 4 local small blocks into the trained network, classifying the 4 local small blocks according to the probability of the network, and averaging to realize the category to which the final classification belongs, thereby achieving the purpose of fingerprint anti-counterfeiting.
Claims (6)
1. An OCT fingerprint anti-counterfeiting method based on resnet50 is characterized by comprising the following steps:
1) the soft rubber material with similar properties to the fingers is utilized: the method comprises the following steps of (1) imitating fingerprint films and fake finger samples with finger fingerprints by using silica gel, gelatin and capacitor gel, and acquiring OCT (optical coherence tomography) volume data of real fingers and fake fingerprint samples by using an OCT (optical coherence tomography) system;
2) filtering and preprocessing data of the slices of different fingerprint samples to obtain smooth data;
3) performing first-order longitudinal differential operation on the slice image, obtaining the position of each row of feature points, performing denoising and fitting on a connected domain to obtain a stratum corneum, selecting a certain feature point on the stratum corneum position, taking local small blocks with the size of 224 x 224 according to 30 upward pixel points of the feature points and 112 left and right pixel points, sequentially spacing 50 pixels, and taking the local small blocks;
4) training and dividing a test set according to the proportion of 8:2 of a data set, constructing a resnet50 network model, setting training parameters and a loss function, and training a resnet50 network model by using the obtained local small blocks with the size of 224 × 224 to obtain a trained resnet50 network model;
5) and (3) randomly selecting feature points in 4B-scan graphs of the data to be identified to obtain local small blocks with the size of 224 x 224, inputting the obtained local small blocks into a trained resnet50 network model, determining the category of the 4 samples according to probability classification of the 4 samples, and performing anti-counterfeiting on the samples.
2. The OCT fingerprint anti-counterfeiting method based on resnet50 as claimed in claim 1, wherein the step 1) of obtaining the fingerprint sample comprises the following steps:
(11) melting the hot melt adhesive, injecting the hot melt adhesive into an experiment platform, slightly pressing a tested finger on the hot melt adhesive when the temperature of the hot melt adhesive is reduced to 42 ℃ and the hot melt adhesive is not solidified, and obtaining a female die;
(12) preparing the prepared soft adhesive material, and injecting the soft adhesive material into the female die to obtain a fingerprint film with the thickness of 0.1-1 mm;
(13) placing the female die in a shaping material similar to a finger, injecting a soft rubber material, and solidifying to obtain an artificial finger sample with a fingerprint;
(14) the obtained fingerprint samples are acquired 3 times in 2 different periods respectively by using an OCT system, and OCT volume data of the fingerprint samples are obtained.
3. The OCT fingerprint anti-counterfeiting method based on resnet50 as claimed in claim 1 or 2, wherein the step 2) comprises the following steps:
(21) filtering the B-scan image acquired by the OCT by using BM3D to remove noise points in the acquisition process;
(22) and carrying out the following preprocessing operations on the filtered image:
where img _ temp is the pixel value of the image, img _ mean is the mean of the image pixels, and img _ std is the standard deviation of the image.
4. The OCT fingerprint anti-counterfeiting method based on resnet50 as claimed in claim 1 or 2, wherein the step 3) comprises the following steps:
(31) by a first longitudinal difference Iy(x, y) to approximate the feature points of the extracted image I:
Iy(x,y)＝I(x+1,y)-I(x,y) (2)
where (x, y) is the pixel coordinate, and I (x, y) represents the gray level of the pixel located at the coordinate (x, y), then each column I is selectedyThe point with the maximum (x, y) value is taken as a characteristic point;
(32) carrying out connected domain denoising on feature points of a 500 x 1800 section image to obtain a position of a finger horny layer, selecting a certain feature point on the position of the horny layer, taking 30 pixels upwards according to the feature points, taking 112 pixels from the left and right, taking local small blocks with the size of 224 x 224, ensuring that the local small blocks taken by training contain horny layer, papillary layer and sweat gland tissue information, and taking the local small blocks at intervals of 50 pixels in sequence to obtain enough training samples;
(33) and (5) repeating the steps (31) and (32), and carrying out OCT slice images on the real finger, the fingerprint film sample and the full finger sample to carry out local small block sampling.
5. The OCT fingerprint anti-counterfeiting method based on resnet50 as claimed in claim 1 or 2, wherein the step 4) comprises the following steps:
(41) taking 80% of the data obtained in the step 3 as a training set and 20% as a test set, wherein the fingerprint images of the real fingers are stored in an img _1 folder, the fingerprint film images are stored in an img _2 folder, and the fake finger images are stored in an img _3 folder;
(42) inputting training samples into a Resnet50 neural network, wherein the Resnet50 network has 50 convolutional layers, 50 ReLU layers, 2 pooling layers, 1 fully-connected layer, and the main structure of Resnet50 is a residual structure, h (x) ═ f (x) + x, if the output of the shallow network is mature enough, i.e. the learning f (x) is equal to 0, this time, it is an identity mapping;
(43) fully connecting the residual module with the features extracted from each convolution layer of the network, specifically classifying the fingerprints by utilizing softmax, using an Adam optimizer as an optimizer, using a cross entropy Loss function as a Loss, loading 4 pictures in a training set into a resnet50 neural network for training, and obtaining the trained network with the iteration number of 100 times;
in the step (43), a cross entropy loss function based on softmax is used; the softmax function is of the form:
in the above formula, ak(x) Representing the output value of the kth class at point x,
representing the sum of the output values of all classes at point x, Pk(x) Represents the probability of class k at point x;
the cross entropy loss function is of the form:
where w (x) represents the weight parameter of the model, l (x) represents the true class of point x, Pl(x)(x) Representing the probability of the true class at point x and E representing the loss value of the cross entropy function.
6. The full convolutional network based training method as claimed in claim 1 or 2, wherein the step (5) comprises the steps of:
(51) randomly selecting 4B-scan images X1, X2, X3 and X4 from the OCT volume data to be identified, and randomly selecting one local small block P1, P2, P3 and P4 from the 4B-scan images according to the step 3);
(52) inputting 4 local small blocks P1, P2, P3 and P4 into a trained network, classifying the probability of the 4 local small blocks, averaging P to (P1+ P2+ P3+ P4)/4, and realizing the category to which the final classification belongs, thereby achieving the purpose of fingerprint anti-counterfeiting.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010293470.7A CN111597895A (en) | 2020-04-15 | 2020-04-15 | OCT fingerprint anti-counterfeiting method based on resnet50 |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010293470.7A CN111597895A (en) | 2020-04-15 | 2020-04-15 | OCT fingerprint anti-counterfeiting method based on resnet50 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111597895A true CN111597895A (en) | 2020-08-28 |
Family
ID=72181362
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010293470.7A Pending CN111597895A (en) | 2020-04-15 | 2020-04-15 | OCT fingerprint anti-counterfeiting method based on resnet50 |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111597895A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117253262A (en) * | 2023-11-15 | 2023-12-19 | 南京信息工程大学 | Fake fingerprint detection method and device based on commonality feature learning |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1124694A (en) * | 1994-12-17 | 1996-06-19 | 欧阳歌军 | Finger print stamp and making method and use thereof |
| CN107480649A (en) * | 2017-08-24 | 2017-12-15 | 浙江工业大学 | A kind of fingerprint pore extracting method based on full convolutional neural networks |
| CN109272507A (en) * | 2018-07-11 | 2019-01-25 | 武汉科技大学 | The layer dividing method of coherent light faultage image based on structure Random Forest model |
| CN109919022A (en) * | 2019-01-29 | 2019-06-21 | 浙江工业大学 | A kind of adaptive inside and outside OCT fingerprint extraction method |
| CN110443217A (en) * | 2019-08-13 | 2019-11-12 | 深圳大学 | One kind being based on multispectral fingerprint method for anti-counterfeit and system |
| CN110610482A (en) * | 2019-08-12 | 2019-12-24 | 浙江工业大学 | Workpiece flaw detection method based on resnet |
-
2020
- 2020-04-15 CN CN202010293470.7A patent/CN111597895A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1124694A (en) * | 1994-12-17 | 1996-06-19 | 欧阳歌军 | Finger print stamp and making method and use thereof |
| CN107480649A (en) * | 2017-08-24 | 2017-12-15 | 浙江工业大学 | A kind of fingerprint pore extracting method based on full convolutional neural networks |
| CN109272507A (en) * | 2018-07-11 | 2019-01-25 | 武汉科技大学 | The layer dividing method of coherent light faultage image based on structure Random Forest model |
| CN109919022A (en) * | 2019-01-29 | 2019-06-21 | 浙江工业大学 | A kind of adaptive inside and outside OCT fingerprint extraction method |
| CN110610482A (en) * | 2019-08-12 | 2019-12-24 | 浙江工业大学 | Workpiece flaw detection method based on resnet |
| CN110443217A (en) * | 2019-08-13 | 2019-11-12 | 深圳大学 | One kind being based on multispectral fingerprint method for anti-counterfeit and system |
Non-Patent Citations (1)
| Title |
|---|
| TARANG CHUGH，ANIL K. JAIN: "OCT Fingerprints: Resilience to Presentation Attacks" * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117253262A (en) * | 2023-11-15 | 2023-12-19 | 南京信息工程大学 | Fake fingerprint detection method and device based on commonality feature learning |
| CN117253262B (en) * | 2023-11-15 | 2024-01-30 | 南京信息工程大学 | Fake fingerprint detection method and device based on commonality feature learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yuan et al. | Fingerprint liveness detection using an improved CNN with image scale equalization | |
| CN104239858B (en) | A kind of method and apparatus of face characteristic checking | |
| Yang et al. | Detecting fake images by identifying potential texture difference | |
| CN112132833B (en) | Dermatological image focus segmentation method based on deep convolutional neural network | |
| CN107145842A (en) | With reference to LBP characteristic patterns and the face identification method of convolutional neural networks | |
| CN109934197A (en) | Training method, device and the computer readable storage medium of human face recognition model | |
| CN108734138B (en) | Melanoma skin disease image classification method based on ensemble learning | |
| CN110084149B (en) | Face verification method based on hard sample quadruple dynamic boundary loss function | |
| CN110298383A (en) | Pathological classification method and system based on multi-modal deep learning | |
| CN108363969B (en) | Newborn pain assessment method based on mobile terminal | |
| CN106295501A (en) | The degree of depth based on lip movement study personal identification method | |
| CN109325513B (en) | Image classification network training method based on massive single-class images | |
| CN112052772A (en) | Face shielding detection algorithm | |
| CN111709914B (en) | Non-reference image quality evaluation method based on HVS characteristics | |
| CN110675335A (en) | Superficial vein enhancement method based on multi-resolution residual error fusion network | |
| CN112069891A (en) | Deep fake face identification method based on illumination characteristics | |
| CN110119695A (en) | A kind of iris activity test method based on Fusion Features and machine learning | |
| CN106709431A (en) | Iris recognition method and device | |
| CN106529395A (en) | Signature image recognition method based on deep brief network and k-means clustering | |
| CN108470178A (en) | A kind of depth map conspicuousness detection method of the combination depth trust evaluation factor | |
| CN105844605B (en) | Based on the human face portrait synthetic method adaptively indicated | |
| CN111597895A (en) | OCT fingerprint anti-counterfeiting method based on resnet50 | |
| CN110348320A (en) | A kind of face method for anti-counterfeit based on the fusion of more Damage degrees | |
| CN111814738A (en) | Human face recognition method, human face recognition device, computer equipment and medium based on artificial intelligence | |
| CN105389573B (en) | A kind of face identification method based on three value mode layering manufactures of part |
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 |