CN106127151B - Based on the finger vein identification method and device for improving local binary model - Google Patents
Based on the finger vein identification method and device for improving local binary model Download PDFInfo
- Publication number
- CN106127151B CN106127151B CN201610463922.5A CN201610463922A CN106127151B CN 106127151 B CN106127151 B CN 106127151B CN 201610463922 A CN201610463922 A CN 201610463922A CN 106127151 B CN106127151 B CN 106127151B
- Authority
- CN
- China
- Prior art keywords
- point
- finger vein
- value
- image
- pixel
- 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
- 210000003462 vein Anatomy 0.000 title claims abstract description 117
- 238000000034 method Methods 0.000 title claims abstract description 39
- 230000004927 fusion Effects 0.000 claims abstract description 10
- 230000009466 transformation Effects 0.000 claims description 25
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims description 23
- 238000007781 pre-processing Methods 0.000 claims description 23
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 9
- 239000013598 vector Substances 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 2
- 239000011159 matrix material Substances 0.000 claims description 2
- 230000010363 phase shift Effects 0.000 claims description 2
- 238000007670 refining Methods 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 102000001554 Hemoglobins Human genes 0.000 description 1
- 108010054147 Hemoglobins Proteins 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000010015 huanglian Substances 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
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
- G06V40/13—Sensors therefor
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Collating Specific Patterns (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The present invention discloses a kind of finger vein identification method and device based on improvement local binary model, can overcome the influence of noise, improves the accuracy of identification and is easy to parallelization realization.S1, finger vein image to be identified is obtained, positioning refers to vein pattern point position, and the LmTP feature of each characteristic point is calculated based on improved local binary model；S2, it is matched by means of finger vein pattern point of the similarity of fusion distance and LmTP feature to the finger vein pattern point and preset template image；S3, according to it is matched as a result, calculate matching double points global similarity；S4, finger hand vein recognition is carried out to the finger vein image according to the global similarity.
Description
Technical Field
The invention relates to the field of image processing and pattern recognition, in particular to a finger vein recognition method and device based on an improved local binary model.
Background
The vein recognition technology is that the oxygen-deprived hemoglobin in the vein blood vessel can absorb near infrared rays with certain wavelength, and the vein blood vessel of a human has good uniqueness. Finger vein recognition attracts attention as a second-generation biometric technology having high forgery prevention properties.
The common finger vein recognition algorithm is a method using local information, such as LBP and LDP, LTP, EQP derived from LBP. The method using the local shape information firstly locates the characteristic points of the finger vein image, and then matches and identifies a new finger vein image through the shapes of the characteristic points. Because only the feature points are used, the template obtained by the algorithm is small, easy to store and small in calculation amount during identification. However, the LBP-like method is sensitive to noise and has a large influence on the recognition accuracy. On the other hand, when matching the feature points, the SVD and hungaran methods are often adopted for matching. The SVD method has the advantages of scale invariance, rotation invariance and the like. However, the disadvantage is that the singular vectors obtained by SVD have negative numbers, and the physical significance of the singular vectors cannot be well explained; on the other hand, the method of SVD decomposition is difficult to parallelize, resulting in a slow speed. The Huanglian method is also called Hungarian algorithm, the algorithm takes the matching problem as the distribution problem, but the algorithm is difficult to converge sometimes, and the optimal solution cannot be found. In addition, when the two methods match the feature points, the probability of the occurrence of mismatching is high.
Disclosure of Invention
Aiming at the defects and shortcomings of the prior art, the invention provides a finger vein identification method and device based on an improved local binary model.
On one hand, the embodiment of the invention provides a finger vein identification method based on an improved local binary model, which comprises the following steps:
s1, acquiring a finger vein image to be identified, positioning the position of the characteristic point of the finger vein, and calculating the LmTP characteristic of each characteristic point based on the improved local binary model;
s2, matching the finger vein feature points with finger vein feature points of a preset template image by means of the similarity of the fusion distance and the LmTP features;
s3, calculating the global similarity of the matching point pairs according to the matching result;
and S4, performing finger vein recognition on the finger vein image according to the global similarity.
On the other hand, an embodiment of the present invention provides a finger vein recognition apparatus based on an improved local binary model, including:
the LmTP feature calculating unit is used for acquiring a finger vein image to be identified, positioning the position of a finger vein feature point and calculating the LmTP feature of each feature point based on an improved local binary model;
the matching unit is used for matching the finger vein feature points with finger vein feature points of a preset template image by means of the similarity of the fusion distance and the LmTP features;
the global similarity calculation unit is used for calculating the global similarity of the matching point pairs according to the matching result;
and the identification unit is used for carrying out finger vein identification on the finger vein image according to the global similarity.
According to the finger vein identification method and device based on the improved local binary model, on one hand, the local binary model is improved, so that the finger vein identification method and device is not easily influenced by noise, on the other hand, the LmTP feature is provided, the influence of the noise can be overcome, and when finger vein feature points are matched, the matching is carried out by means of the similarity of the fusion distance and the LmTP feature, an optimal matching result can be obtained, so that when the number of matching points is few, the mismatching situation is not easy to occur, the matching accuracy is higher, and the parallelization is easy to realize.
Drawings
FIG. 1 is a schematic flowchart of an embodiment of a finger vein recognition method based on an improved local binary model according to the present invention;
FIG. 2 is a schematic flow chart of one embodiment of S2 in FIG. 1;
fig. 3 is a schematic structural diagram of an embodiment of the finger vein recognition apparatus based on an improved local binary model according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some embodiments, but not all embodiments, of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, the present embodiment discloses a finger vein identification method based on an improved local binary model, including:
s1, acquiring a finger vein image to be identified, positioning the position of the characteristic point of the finger vein, and calculating the LmTP characteristic of each characteristic point based on the improved local binary model;
s2, matching the finger vein feature points with finger vein feature points of a preset template image by means of the similarity of the fusion distance and the LmTP features;
s3, calculating the global similarity of the matching point pairs according to the matching result;
and S4, performing finger vein recognition on the finger vein image according to the global similarity.
According to the finger vein identification method based on the improved local binary model, on one hand, the local binary model is improved, so that the finger vein identification method is not easily influenced by noise, on the other hand, the LmTP feature is provided, the LmTP feature can overcome the influence of the noise, and when finger vein feature points are matched, the matching is carried out by means of the similarity of the fusion distance and the LmTP feature, so that the optimal matching result can be obtained, the mismatching condition is not easily caused when the matching points are few, the matching accuracy is higher, the parallelization is easy to realize, and compared with the prior art, the method can overcome the influence of the noise, improve the identification accuracy and is easy to realize parallelization.
Optionally, in another embodiment of the finger vein identification method based on an improved local binary model, the locating the position of the finger vein feature point includes:
preprocessing the finger vein image;
identifying a finger vein region and a background region in the image obtained by preprocessing by using a maximum curvature algorithm, carrying out binarization on the image obtained by preprocessing to obtain a binary image, and refining the binary image, wherein the pixel value of the finger vein region is 1, and the pixel value of the background region is 0;
traversing each position of the thinned binary image by taking a window of 3 x 3, judging whether the pixel value of a pixel point at the central point of the window is 1, and if so, calculating NtransJudgment of NtransWhether it is 2 or not less than 6,if N is presenttrans2, the pixel Point at the position is an End Point (EP), or if N istransIf the window is not less than 6, the pixel Point at the position is a Branch Point (BP), wherein the center Point of the window is located at the position, N istransIs calculated by the formula
In the formula, p1To p8The pixel values p of the pixel points which are distributed clockwise along the periphery of the window from the upper left corner position in the window are sequentially9＝p1。
In the embodiment of the invention, the difference between the finger vein area and the background area can be enhanced while noise is smoothed by preprocessing the finger vein image.
Optionally, in another embodiment of the finger vein identification method based on an improved local binary model, the preprocessing the finger vein image includes:
filtering the finger vein image using an even symmetric Gabor filter, wherein the Gabor filter function is expressed as
Wherein, x 'is xcos θ + ysin θ, y' is xsin θ + ycos θ, (x, y) is the original coordinate, (x ', y') is the filtered coordinate, and λ is the wavelength; theta is the direction, and the parameter controls the stripe direction of the Gabor function and takes the value of 0-360 degrees;the phase shift is carried out, and the value range is-180 degrees to-180 degrees; gamma is the length-width ratio, determines the ellipticity of the Gabor function, and sigma represents the standard deviation of the Gaussian factor of the Gabor function;
the filtered image is enhanced using CLAHE.
In the embodiment of the present invention, the enhancement of the filtered image by CLAHE is the prior art, and is not described herein again, for example, the method in the patent with the application number 201510661549.X may be adopted.
Optionally, in another embodiment of the method for identifying a finger vein based on an improved local binary model according to the present invention, the identifying a finger vein region and a background region in the pre-processed image by using a maximum curvature algorithm includes:
calculating the curvature of each pixel point in the image obtained by the preprocessing by the formulaWherein, Kf(z) represents the curvature of a pixel point located at z,(x0,y0) As the coordinate of position z, when i>At 0, xiThe abscissa of the ith pixel point at the right side of the pixel point at the position z is expressed, when i is<At 0, xiThe abscissa of the ith pixel point at the left side of the pixel point at the position z is expressed when i is>At 0, yiThe ordinate of the ith pixel point right above the pixel point at the position z is expressed when i is<At 0, yiThe vertical coordinate of the ith pixel point right below the pixel point at the position z is represented, w is a parameter and is used for calculating the average value of the section, and the empirical value of the parameter is 8;
determining a maximum value of the curvature, and taking a pixel point corresponding to the maximum value as a possible finger vein central point;
scoring each possible finger vein center point, and calculating the score according to the formula of S'cr＝Kf×WrWherein, S'crIs a first score, KfIs a curvature, WrThe width of the region with positive local curvature of the central point of the finger vein;
for each pixel point in the image obtained by preprocessing, summing first scores of possible finger vein center points in the horizontal direction, the vertical direction and the +/-45-degree direction, and taking the summation result as a second score of the pixel point;
for each pixel point in the image obtained by preprocessing, respectively calculating C of the pixel point in the horizontal direction, the vertical direction and the +/-45-degree directiond(x, y) and C for the pixel point in the horizontal direction, the vertical direction, and the + -45 DEG directiond(x, y) obtaining the maximum value to obtain G (x, y) of the pixel point, wherein,
Cd(x,y)＝min{max(Scr(x-1,y),Scr(x-2,y)),max(Scr(x+1,y),Scr(x +2, y)) }, (x, y) denotes the coordinates of the pixel point, Scr(x-1, y) is the second score of the first pixel point before the pixel point with the coordinate (x, y) in a certain direction, Scr(x-2, y) is the second score of the second pixel before the pixel with coordinate (x, y) in a certain direction, Scr(x +1, y) is a second score of a first pixel point behind a pixel point with coordinates (x, y) in a certain direction, and Scr(x +2, y) is a second score of a second pixel point behind the pixel point with the coordinate (x, y) in a certain direction;
and for each pixel point in the image obtained by preprocessing, comparing the magnitude relation between G (x, y) of the pixel point and a first numerical value, if the G (x, y) of the pixel point is smaller than the first numerical value, determining that the pixel point belongs to a background area, and otherwise, determining that the pixel point belongs to a finger vein area.
In the embodiment of the invention, the possible finger vein central points in the horizontal direction and the vertical direction respectively pass through the possible finger vein central points in the horizontal direction straight line and the vertical direction straight line of the pixel point, the possible finger vein central points in the + 45-degree direction and the-45-degree direction respectively pass through the pixel point, and the slope is 1 or-1. For the first numerical value, the values greater than 0 in G (x, y) may be sorted from large to small, and the value of the sorted middle element is taken as the first numerical value, or other values may be taken as needed, for example, other values near the value of the middle element, which is not limited in the present invention.
Optionally, in another embodiment of the finger vein recognition method based on an improved local binary model of the present invention, the calculating the LmTP feature of each feature point based on the improved local binary model includes:
for each feature point, calculatingAndwill be described in Andconnected to get the LmTP feature of the feature point, where,
a0to a7The gray values of pixel points which are distributed clockwise along the periphery of the window from the upper left corner position in a 3-by-3 window which takes the feature point as the center in the image obtained by the preprocessing, b0To b15The gray value c of pixel points which are distributed clockwise along the periphery of the window from the upper left corner position in a 5-by-5 window which takes the feature point as the center in the image obtained by the preprocessing are sequentially set0To c23Sequentially starting from the upper left corner position in a 7-7 window which takes the feature point as the center in the image obtained by the preprocessing, and the gray values of pixel points which are distributed clockwise along the periphery of the window, i belongs to (1,2, …,7), m is the gray value of the feature point,
t is a predetermined value, typically 5.
Alternatively, referring to fig. 2, in another embodiment of the finger vein identification method based on an improved local binary model of the present invention, the S2 includes:
s20, judging whether the current cycle number reaches a second value;
s21, if the second value is reached, saving the best matching point pairAnd the optimal affine transformation parameters S, T, performing step S32, otherwise, performing step S22;
s22, determining affine transformation parameters from the finger vein image to the template image according to the end points of the finger vein image and the end points of the template image;
s23, judging whether the affine transformation parameters are in a preset range, if so, executing a step S24, otherwise, executing a step S20;
s24, affine transformation is carried out on the characteristic points of the finger vein image through affine transformation parameters, and the positions of the transformed characteristic points are expressed as (f)xi',fyi')，i＝1,…,n1，n1The number of the characteristic points of the finger vein image is obtained;
s25, calculation using distance information (f)xi',fyi') and (g)xj,gyj) Wherein, (g)xj,gyj) J is 1, …, n is the position of the characteristic point of the template image2，n2The number of the characteristic points of the template image is obtained;
s26, calculating the local similarity based on the distance of the current matching point pair S1;
s27, judging whether the local similarity based on the distance S1 is smaller than a third numerical value;
s28, if the value is less than the third value, updating S to the affine transformation parameter and updating the best matching point pairUpdating the third value to be S1 for the best matching point pair, and executing step S29, otherwise, executing step S29 directly,dlthe Euclidean distance of the ith best matching point pair is defined, and M is the number of the best matching point pairs;
s29, calculating local Similarity of the current matching point pair based on LmTP characteristicsL；
S30, judging the local Similarity based on the LmTP characteristicsLWhether it is greater than a fourth value;
s31, if the value is larger than the fourth value, updating T to the affine transformation parameter and updating the best matching point pairUpdating the fourth value to be Similarity for the best matching point pairLAnd performing the step S20, otherwise, directly performing the step S20;
s32, affine transformation is carried out on the branch points of the finger vein image by the calculated optimal affine transformation parameters S, T, the branch points after each transformation are matched with the branch points of the template image, and matching point pairs are obtained
In the embodiment of the present invention, the second value may be set as needed, for example, may be set to 50, 60, and so on. For the case that the finger vein image translation, rotation and scaling exist, the following formula in the form of homogeneous coordinates can be used for calculating the affine transformation parameter p1-p6：
Wherein { (x)i,yi) I ═ 1,2,3} are coordinates of three endpoints of the finger vein image, { (x)i',yi') | i ═ 1,2,3} is the position of the three end points of the template image, where (x) isi,yi) Corresponds to (x)i',yi'). Through three groups of corresponding coordinate information, a group of affine transformation parameters p can be obtained1-p6. The preset range at this time means:
0.5≤p1≤1.5，
-0.5≤p2≤0.5，
-40≤p3≤40，
-0.5≤p4≤0.5，
0.5≤p5≤1.5，
-40≤p6≤40。
for the case of the presence of translation, rotation, scaling parts, the affine transformation parameter is p1-p6Some of the parameters in (1) are not described in detail herein. When the optimal affine transformation parameters are calculated for the feature points, the end points are used, and the calculated affine transformation parameters are more accurate because the end points are more than branch points.
The initial value of the third value may be a larger value, such as 104、1.7×10308And the like. The initial value of the fourth value is typically 0.
Optionally, in another embodiment of the finger vein identification method based on the improved local binary model, the local Similarity based on the LmTP feature of the current matching point pair is calculatedLThe method comprises the following steps:
calculating Euclidean distance d between the transformed ith characteristic point and the jth characteristic point of the template imageijThe calculation formula is
dij＝((fxi'-gxj)2+(fyi'-gyj)2)0.5；
A distance matrix D is calculated in which,
wherein D isij＝exp(-dij)；
And determining the characteristic point pair corresponding to the element value which is the maximum in the affiliated row and affiliated column in the D as the matched characteristic point pair, wherein the matched characteristic point pair is simply M1 { (f)n',gn) N is 1, …, N is the total number of matching point pairs;
calculating SimilarityLThe calculation formula is
Wherein,Ln1(k) and Ln2(k) The kth bit of the LmTP feature of two feature points in the nth matching point pair, Len is 64.
Feature L for a pair of LmTPi1And Li2The length is 64, and because the LmTP feature contains-1, the similarity can not be calculated by using the Hamming distance, so the similarity Sim of the pair of features is calculated by the following formulaG(i)：
If N characteristic points are matched, the Similarity of the matched point pairsGCan be calculated by the following formula:
similarityGThe smaller the size, the more similar the two templates are represented. However, the similarity may cause mismatching, for example, when the matched feature points are few, the global similarity value may be small, which may cause mismatching. Thus, the above-described similarity is referred to as a global similarity. At the same time, the invention modifies the global similarity, Li1And Li2The local similarity of the pair of features is calculated by:
local Similarity of matching point pairsLCan be calculated by:
the greater the local similarity, the more similar the two templates. After the local similarity is used, when the matched characteristic point pairs are fewer, the local similarity is not too large, so that mismatching is avoided.
Optionally, in another embodiment of the finger vein identification method based on an improved local binary model of the present invention, the S3 includes:
computing pairs of best matching pointsGlobal similarity ofAnd global similarity ofAnd selecting the smaller one as the global similarity of the matching point pair.
Optionally, in another embodiment of the finger vein identification method based on an improved local binary model of the present invention, the S4 includes:
and comparing the global similarity with a fifth numerical value, if the global similarity is greater than the fifth numerical value, determining the finger to be different, and otherwise, determining the finger to be the same.
In the embodiment of the present invention, the fifth numerical value is generally 0.25.
Referring to fig. 3, the present embodiment discloses a finger vein recognition apparatus based on an improved local binary model, including:
the LmTP feature calculating unit 1 is used for acquiring a finger vein image to be identified, positioning the position of a finger vein feature point, and calculating LmTP features of each feature point based on an improved local binary model;
a matching unit 2, configured to match the finger vein feature points with finger vein feature points of a preset template image by means of a fusion distance and a similarity of LmTP features;
the global similarity calculation unit 3 is used for calculating the global similarity of the matching point pairs according to the matching result;
and the identification unit 4 is used for carrying out finger vein identification on the finger vein image according to the global similarity.
The finger vein recognition device based on the improved local binary model provided by the embodiment of the invention has the advantages that on one hand, the local binary model is improved, so that the finger vein recognition device is not easily influenced by noise, on the other hand, the LmTP feature is provided, the influence of the noise can be overcome, and when finger vein feature points are matched, the matching is carried out by means of the similarity of the fusion distance and the LmTP feature, so that the optimal matching result can be obtained, the mismatching condition is not easily caused when the matching points are few, the matching accuracy is higher, the parallelization is easy to realize, and compared with the prior art, the influence of the noise can be overcome, the recognition accuracy is improved, and the parallelization is easy to realize.
Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations fall within the scope defined by the appended claims.
Claims (8)
1. A finger vein identification method based on an improved local binary model is characterized by comprising the following steps:
s1, acquiring a finger vein image to be identified, positioning the position of the characteristic point of the finger vein, and calculating the LmTP characteristic of each characteristic point based on the improved local binary model;
s2, matching the finger vein feature points with finger vein feature points of a preset template image by means of the similarity of the fusion distance and the LmTP features;
s3, calculating the global similarity of the matching point pairs according to the matching result;
s4, carrying out finger vein recognition on the finger vein image according to the global similarity;
wherein the computing LmTP features of the respective feature points based on the improved local binary model comprises:
for each feature point, calculatingAndwill be described in Andconnected to obtain a row vector as the LmTP feature of the feature point, wherein,
a0to a7The gray values of pixel points which are distributed clockwise along the periphery of the window from the upper left corner position in a 3-by-3 window which takes the feature point as the center in the image obtained by the preprocessing, b0To b15The gray value c of pixel points which are distributed clockwise along the periphery of the window from the upper left corner position in a 5-by-5 window which takes the feature point as the center in the image obtained by the preprocessing are sequentially set0To c23Sequentially starting from the upper left corner position in a 7-7 window which takes the feature point as the center in the image obtained by the preprocessing, and the gray values of pixel points which are distributed clockwise along the periphery of the window are i epsilon (1,2, …,7), m is the gray value of the feature point, wherein "&"is defined as follows
t is a preset numerical value and takes the value of 5;
wherein the S2 includes:
s20, judging whether the current cycle number reaches a second value;
s21, if the second value is reached, saving the best matching point pairAnd the optimal affine transformation parameters S, T, performing step S32, otherwise, performing step S22;
s22, determining affine transformation parameters from the finger vein image to the template image according to the end points of the finger vein image and the end points of the template image;
s23, judging whether the affine transformation parameters are in a preset range, if so, executing a step S24, otherwise, executing a step S20;
s24, affine transformation is carried out on the feature points of the finger vein image through affine transformation parameters, and the positions of the transformed feature points are recorded as (f'xi,f′yi)，i＝1,…,n1，n1The number of the characteristic points of the finger vein image is obtained;
s25, calculating (f ') by using distance information'xi,f′yi) And (g)xj,gyj) Wherein, (g)xj,gyj) J is 1, …, n is the position of the characteristic point of the template image2，n2The number of the characteristic points of the template image is obtained;
s26, calculating the local similarity based on the distance of the current matching point pair S1;
s27, judging whether the local similarity based on the distance S1 is smaller than a third numerical value;
s28, if the value is less than the third value, updating S to the affine transformation parameter and updating the best matching point pairUpdating the third value to be S1 for the best matching point pair, and executing step S29, otherwise, executing step S29 directly,dlthe Euclidean distance of the ith best matching point pair is defined, and M is the number of the best matching point pairs;
s29, calculating local Similarity of the current matching point pair based on LmTP characteristicsL；
S30, judging the local Similarity based on the LmTP characteristicsLWhether it is greater than a fourth value;
s31, if the value is larger than the fourth value, updating T to the affine transformation parameter and updating the best matching point pairUpdating the fourth value to be Similarity for the best matching point pairLAnd performing the step S20, otherwise, directly performing the step S20;
s32, affine transformation is carried out on the branch points of the finger vein image by using the calculated optimal affine transformation parameters S, T, and each transformation is carried outMatching the branch point with the branch point of the template image to obtain a matching point pair
2. The method of claim 1, wherein said locating a finger vein landmark position comprises:
preprocessing the finger vein image;
identifying a finger vein region and a background region in the image obtained by preprocessing by using a maximum curvature algorithm, carrying out binarization on the image obtained by preprocessing to obtain a binary image, and refining the binary image, wherein the pixel value of the finger vein region is 1, and the pixel value of the background region is 0;
traversing each position of the thinned binary image by taking a window of 3 x 3, judging whether the pixel value of a pixel point at the central point of the window is 1, and if so, calculating NtransJudgment of NtransWhether it is 2 or not less than 6, if Ntrans2, the pixel point at the position is the end point, or if N istransIf the window is not less than 6, the pixel point at the position is a branch point, wherein the center point of the window is located at the position, NtransIs calculated by the formula
In the formula, p1To p8The pixel values p of the pixel points which are distributed clockwise along the periphery of the window from the upper left corner position in the window are sequentially9＝p1。
3. The method of claim 2, wherein the pre-processing the finger vein image comprises:
filtering the finger vein image using an even symmetric Gabor filter, wherein the Gabor filter function is expressed as
Wherein, x 'is xcos θ + ysin θ, y' is xsin θ + ycos θ, (x, y) is the original coordinate, (x ', y') is the filtered coordinate, and λ is the wavelength; theta is the direction, and the parameter controls the stripe direction of the Gabor function and takes the value of 0-360 degrees;the phase shift is carried out, and the value range is-180 degrees to-180 degrees; gamma is the length-width ratio, determines the ellipticity of the Gabor function, and sigma represents the standard deviation of the Gaussian factor of the Gabor function;
the filtered image is enhanced using CLAHE.
4. The method of claim 3, wherein the identifying finger vein regions and background regions in the pre-processed image using a maximum curvature algorithm comprises:
calculating the curvature of each pixel point in the image obtained by the preprocessing by the formulaWherein, Kf(z) represents the curvature of a pixel point located at z,(x0,y0) Is the coordinate of position z, when i > 0, xiThe abscissa of the ith pixel point at the right side of the pixel point at the position z is represented, and when i is less than 0, xiThe abscissa of the ith pixel point at the left side of the pixel point at the position z is represented, and when i is greater than 0, yiThe ordinate of the ith pixel point right above the pixel point at the position z is represented, and when i is less than 0, yiThe ordinate of the ith pixel point directly below the pixel point at the position z is represented, and w is a parameter used for calculating the average value of the section and the experience of the sectionA value of 8;
determining a maximum value of the curvature, and taking a pixel point corresponding to the maximum value as a possible finger vein central point;
scoring the center point of each possible finger vein, and calculating the score according to the formula Sc'r＝Kf×WrWherein, S'crIs a first score, KfIs a curvature, WrThe width of the region with positive local curvature of the central point of the finger vein;
for each pixel point in the image obtained by preprocessing, summing first scores of possible finger vein center points in the horizontal direction, the vertical direction and the +/-45-degree direction, and taking the summation result as a second score of the pixel point;
for each pixel point in the image obtained by preprocessing, respectively calculating C of the pixel point in the horizontal direction, the vertical direction and the +/-45-degree directiond(x, y) and C for the pixel point in the horizontal direction, the vertical direction, and the + -45 DEG directiond(x, y) obtaining the maximum value to obtain G (x, y) of the pixel point, wherein,
Cd(x,y)＝min{max(Scr(x-1,y),Scr(x-2,y)),max(Scr(x+1,y),Scr(x +2, y)) }, (x, y) denotes the coordinates of the pixel point, Scr(x-1, y) is the second score of the first pixel point before the pixel point with the coordinate (x, y) in a certain direction, Scr(x-2, y) is the second score of the second pixel before the pixel with coordinate (x, y) in a certain direction, Scr(x +1, y) is a second score of a first pixel point behind a pixel point with coordinates (x, y) in a certain direction, and Scr(x +2, y) is a second score of a second pixel point behind the pixel point with the coordinate (x, y) in a certain direction;
and for each pixel point in the image obtained by preprocessing, comparing the magnitude relation between G (x, y) of the pixel point and a first numerical value, if the G (x, y) of the pixel point is smaller than the first numerical value, determining that the pixel point belongs to a background area, and otherwise, determining that the pixel point belongs to a finger vein area.
5. The method of claim 1, wherein the computing the local Similarity based on the LmTP feature for the current matching point pair is performed based onLThe method comprises the following steps:
calculating Euclidean distance d between the transformed ith characteristic point and the jth characteristic point of the template imageijThe calculation formula is
dij＝((f′xi-gxj)2+(f′yi-gyj)2)0.5；
A distance matrix D is calculated in which,
wherein D isij＝exp(-dij)；
And determining the characteristic point pair corresponding to the element value with the maximum value in the affiliated row and affiliated column in D as a matched characteristic point pair, wherein the matched characteristic point pair is simply M1 { (f'n,gn) N is 1, …, N is the total number of matching point pairs;
calculating SimilarityLThe calculation formula is
Wherein,Ln1(k) and Ln2(k) The kth bit of the LmTP feature of two feature points in the nth matching point pair, Len is 64.
6. The method according to claim 1, wherein the S3 includes:
computing pairs of best matching pointsGlobal similarity ofAnd global similarity ofAnd selecting the smaller one as the global similarity of the matching point pair.
7. The method according to claim 1, wherein the S4 includes:
and comparing the global similarity with a fifth numerical value, if the global similarity is greater than the fifth numerical value, determining the finger to be different, and otherwise, determining the finger to be the same.
8. A finger vein recognition device based on an improved local binary model is characterized by comprising:
the LmTP feature calculating unit is used for acquiring a finger vein image to be identified, positioning the position of a finger vein feature point and calculating the LmTP feature of each feature point based on an improved local binary model;
the matching unit is used for matching the finger vein feature points with finger vein feature points of a preset template image by means of the similarity of the fusion distance and the LmTP features;
the global similarity calculation unit is used for calculating the global similarity of the matching point pairs according to the matching result;
and the identification unit is used for carrying out finger vein identification on the finger vein image according to the global similarity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610463922.5A CN106127151B (en) | 2016-06-23 | 2016-06-23 | Based on the finger vein identification method and device for improving local binary model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610463922.5A CN106127151B (en) | 2016-06-23 | 2016-06-23 | Based on the finger vein identification method and device for improving local binary model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106127151A CN106127151A (en) | 2016-11-16 |
| CN106127151B true CN106127151B (en) | 2019-04-19 |
Family
ID=57268070
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610463922.5A Active CN106127151B (en) | 2016-06-23 | 2016-06-23 | Based on the finger vein identification method and device for improving local binary model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106127151B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107229899B (en) * | 2017-03-24 | 2020-07-24 | 杭州电子科技大学 | Finger vein identification method based on two-dimensional Gaussian maximum curvature |
| CN107704822A (en) * | 2017-09-30 | 2018-02-16 | 广州智慧城市发展研究院 | The extraction of finger vein further feature and matching process and system based on incomplete completion |
| CN110443128B (en) * | 2019-06-28 | 2022-12-27 | 广州中国科学院先进技术研究所 | Finger vein identification method based on SURF feature point accurate matching |
| CN111222456B (en) * | 2020-01-04 | 2023-06-23 | 圣点世纪科技股份有限公司 | High-speed retrieval algorithm under ultra-large user quantity of finger veins |
| CN111382703B (en) * | 2020-03-10 | 2023-06-23 | 大连海事大学 | Finger vein recognition method based on secondary screening and score fusion |
| CN111639555B (en) * | 2020-05-15 | 2023-06-20 | 圣点世纪科技股份有限公司 | Finger vein image noise accurate extraction and adaptive filtering denoising method and device |
| CN117292413B (en) * | 2023-11-27 | 2024-05-07 | 江苏圣点世纪科技有限公司 | Vein image rotation identification method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254165A (en) * | 2011-08-12 | 2011-11-23 | 北方工业大学 | Hand back vein identification method based on fusion of structural coding features and texture coding features |
| CN102411711A (en) * | 2012-01-04 | 2012-04-11 | 山东大学 | Finger vein recognition method based on individualized weight |
| CN103235929A (en) * | 2013-03-11 | 2013-08-07 | 北京航空航天大学 | Identification method and identification device on basis of hand vein images |
| KR101450247B1 (en) * | 2012-12-31 | 2014-10-15 | 기초과학연구원 | Method for authenticating based on finger vein using SIFT keypoint |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4407714B2 (en) * | 2007-04-06 | 2010-02-03 | セイコーエプソン株式会社 | Biometric authentication device and biometric authentication method |
-
2016
- 2016-06-23 CN CN201610463922.5A patent/CN106127151B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254165A (en) * | 2011-08-12 | 2011-11-23 | 北方工业大学 | Hand back vein identification method based on fusion of structural coding features and texture coding features |
| CN102411711A (en) * | 2012-01-04 | 2012-04-11 | 山东大学 | Finger vein recognition method based on individualized weight |
| KR101450247B1 (en) * | 2012-12-31 | 2014-10-15 | 기초과학연구원 | Method for authenticating based on finger vein using SIFT keypoint |
| CN103235929A (en) * | 2013-03-11 | 2013-08-07 | 北京航空航天大学 | Identification method and identification device on basis of hand vein images |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106127151A (en) | 2016-11-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106127151B (en) | Based on the finger vein identification method and device for improving local binary model | |
| CN105447441B (en) | Face authentication method and device | |
| CN106096569B (en) | A kind of finger vein identification method | |
| US20120057763A1 (en) | method for recognizing the identity of user by biometrics of palm vein | |
| CN103955674B (en) | Palm print image acquisition device and palm print image positioning and segmenting method | |
| CN106127193B (en) | A kind of facial image recognition method | |
| CN104239769A (en) | Identity recognition method and system based on finger vein characteristics | |
| CN112597812A (en) | Finger vein identification method and system based on convolutional neural network and SIFT algorithm | |
| Manana et al. | Preprocessed faster RCNN for vehicle detection | |
| CN106203375A (en) | A kind of based on face in facial image with the pupil positioning method of human eye detection | |
| CN109977909B (en) | Finger vein identification method and system based on minutia area matching | |
| WO2018176514A1 (en) | Fingerprint registration method and device | |
| CN108229247A (en) | A kind of mobile vehicle detection method | |
| Wang et al. | Damaged fingerprint classification by Deep Learning with fuzzy feature points | |
| Lu et al. | Contour box: Rejecting object proposals without explicit closed contours | |
| CN110147769B (en) | Finger vein image matching method | |
| Derman et al. | Normalized cross-correlation based global distortion correction in fingerprint image matching | |
| Podder et al. | An efficient iris segmentation model based on eyelids and eyelashes detection in iris recognition system | |
| Huang et al. | Robust finger vein recognition based on deep CNN with spatial attention and bias field correction | |
| CN113936303B (en) | Method for determining maximum inscribed rectangle of hand image and image identification method | |
| Shen et al. | CNN-based high-resolution fingerprint image enhancement for pore detection and matching | |
| Chen et al. | Band selection for improvement of dorsal hand recognition | |
| CN112926516B (en) | Robust finger vein image region-of-interest extraction method | |
| CN114973308A (en) | Finger vein identification method and system based on elastic weight solidification and multivariate similarity loss | |
| WO2016004706A1 (en) | Method for improving iris recognition performance in non-ideal environment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230215 Address after: 100070 unit 7, 3 / F, building 4, District 4, 186 South 4th Ring Road West, Fengtai District, Beijing (Park) Patentee after: Beijing Haixin Zhisheng Technology Co.,Ltd. Address before: 100070 Beijing city Fengtai District Haiying Road No. 6 Building No. 29 hospital Patentee before: Beijing Haixin Kejin High-Tech Co.,Ltd. |