CN109101867A - A kind of image matching method, device, computer equipment and storage medium - Google Patents
A kind of image matching method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN109101867A CN109101867A CN201810593681.5A CN201810593681A CN109101867A CN 109101867 A CN109101867 A CN 109101867A CN 201810593681 A CN201810593681 A CN 201810593681A CN 109101867 A CN109101867 A CN 109101867A
- Authority
- CN
- China
- Prior art keywords
- image
- pixel
- legal
- characteristic point
- point
- 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.)
- Withdrawn
Links
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
- G06V40/1365—Matching; Classification
- G06V40/1376—Matching features related to ridge properties or fingerprint texture
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)
Abstract
The present invention relates to field of computer technology, providing a kind of image matching method, device, computer equipment and storage medium, image matching method includes: to obtain finger venous image to be detected and carry out gray processing processing, obtains gray level image；The characteristic point of finger vein grain in gray level image is extracted using fast algorithm；Characteristic point is handled using sift algorithm, obtains the feature vector of each characteristic point；For each characteristic point, the cosine value between the feature vector of this feature point and each of the legal image feature vector of legal characteristic point is calculated；If cosine value is less than preset cosine threshold value, the characteristic point for obtaining the cosine value is confirmed as match point；If the quantity of match point is greater than preset points threshold value, confirm that finger venous image to be detected matches with legal image.Technical solution of the present invention realizes accurately identifying and matching to finger vein grain, to effectively improve the matched accuracy of finger vein grain.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of image matching method, device, computer equipment and deposit
Storage media
Background technique
Currently, traditional finger vena matching algorithm matching effect is undesirable, there are false recognition rate and complexity is realized
It is larger, cause matched accuracy not high, so that in the link of finger vein grain verifying, since finger vein grain matches
Accuracy it is not high, be frequently present of authentication failed situation, reduction is proved to be successful rate.
Summary of the invention
Based on this, it is necessary to which in view of the above technical problems, providing one kind, that finger vein grain can be improved is matched accurate
Image matching method, device, computer equipment and the storage medium of property.
A kind of image matching method, comprising:
Obtain finger venous image to be detected；
Gray processing processing is carried out to the finger venous image to be detected, obtains gray level image；
The characteristic point of finger vein grain in the gray level image is extracted using fast algorithm；
The characteristic point is handled using sift algorithm, obtains the feature vector of each characteristic point；
For each characteristic point, the feature vector and the legal feature of each of legal image of the characteristic point are calculated
Cosine value between the feature vector of point, wherein the legal image refers to the pre- finger venous image for first passing through legitimate authentication；
If the cosine value is less than preset cosine threshold value, the characteristic point for obtaining the cosine value is confirmed as match point；
If the quantity of the match point be greater than preset points threshold value, confirm the finger venous image to be detected with
The legal image matches.
A kind of image matching apparatus, comprising:
Module is obtained, for obtaining finger venous image to be detected；
Gray scale module obtains gray level image for carrying out gray processing processing to the finger venous image to be detected；
Feature point extraction module, for extracting the feature of finger vein grain in the gray level image using fast algorithm
Point；
Vector calculation module obtains each characteristic point for handling using sift algorithm the characteristic point
Feature vector；
Similarity calculation module, for being directed to each characteristic point, calculate the feature vector of the characteristic point with it is legal
Cosine value between the feature vector of the legal characteristic point of each of image, wherein the legal image, which refers to, pre- first passes through conjunction
The finger venous image of method certification；
Identification module will obtain the characteristic point of the cosine value if being less than preset cosine threshold value for the cosine value
It is confirmed as match point；
First matching module confirms described to be checked if the quantity for the match point is greater than preset points threshold value
The finger venous image of survey matches with the legal image.
A kind of computer equipment, including memory, processor and storage are in the memory and can be in the processing
The computer program run on device, the processor realize the step of above-mentioned image matching method when executing the computer program
Suddenly.A kind of computer readable storage medium, the computer-readable recording medium storage have computer program, the computer journey
The step of above-mentioned image matching method is realized when sequence is executed by processor.
Above-mentioned image matching method, device, computer equipment and storage medium, by finger venous image to be detected
Gray processing processing is carried out, gray level image is obtained, the feature of finger vein grain in gray level image is extracted using fast algorithm
Point obtains the feature vector of each characteristic point by sift algorithm, and according to feature vector, calculates the characteristic point in gray level image
Feature vector and each of the legal image feature vector of legal characteristic point between cosine value, judge further according to cosine value
The number of match point is finally compared by the number of match point with preset threshold, if the number of match point is greater than preset threshold,
Then confirm that finger venous image to be detected matches with legal image.On the one hand, finger vena line is extracted using fast algorithm
The characteristic point on road realizes the rapidly extracting to the characteristic point of finger vein grain, to improve using the high efficiency of fast algorithm
To the recognition efficiency of finger vein grain characteristic point；On the other hand, it is obtained in such a way that fast algorithm and sift algorithm combine
To the feature vector of characteristic point, and the cosine value by calculating feature vector determines match point, realizes to finger vein grain
Accurate match, effectively improve the matched accuracy of finger vein grain.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below by institute in the description to the embodiment of the present invention
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the invention
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is the application environment schematic diagram of images match reason method provided in an embodiment of the present invention；
Fig. 2 is the flow chart of image matching method provided in an embodiment of the present invention；
Fig. 3 is the flow chart of step S2 in image matching method provided in an embodiment of the present invention；
Fig. 4 is the flow chart of step S3 in image matching method provided in an embodiment of the present invention；
Fig. 5 is the flow chart of step S4 in image matching method provided in an embodiment of the present invention；
Fig. 6 be in image matching method provided in an embodiment of the present invention displacement calculate by way of carry out again it is matched
Flow chart；
Fig. 7 is the schematic diagram of image matching apparatus provided in an embodiment of the present invention；
Fig. 8 is the schematic diagram of computer equipment provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are some of the embodiments of the present invention, instead of all the embodiments.Based on this hair
Embodiment in bright, every other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, shall fall within the protection scope of the present invention.
Fig. 1 shows application environment provided in an embodiment of the present invention, which includes server-side and client,
In, it is attached between server-side and client by network, client is acquired for finger vein image, and will
Collected finger venous image is sent to server-side, client specifically can be, but not limited to be video camera, camera, scanner or
Person has the finger vein image acquisition equipment of other camera functions；Server-side carries out finger vena for finger vein image
Lines matching, server-side can specifically be realized with the server cluster that independent server or multiple servers form.This hair
The image matching method that bright embodiment provides is applied to server-side.
In one embodiment, as shown in Fig. 2, providing a kind of image matching method, the service in Fig. 1 is applied in this way
It is illustrated, includes the following steps: for end
S1: finger venous image to be detected is obtained.
In embodiments of the present invention, finger venous image to be detected refers to without any processing, directly from finger vein
Acquire collected finger venous image in equipment.
S2: gray processing processing is carried out to finger venous image to be detected, obtains gray level image.
In embodiments of the present invention, gray processing refers in RGB model, if when R=G=B, color representation only has one
Kind greyscale color, wherein the value of R=G=B is gray value, and therefore, each pixel of gray level image only needs a byte storage gray scale
Value, tonal range 0-255.
By carrying out gray processing processing to finger venous image to be detected, by all pixels point in finger venous image
Pixel value is all adjusted between 0-255, obtains the gray level image of finger venous image.
S3: the characteristic point of finger vein grain in gray level image is extracted using fast algorithm.
In embodiments of the present invention, characteristic point refers to the pixel that can be identified for that the finger vein grain in gray level image,
By using Corner Detection (Features from accelerated segment test, fast) algorithm, step S2 is obtained
To gray level image handled, extract the pixel of finger vein grain in the gray level image as characteristic point.
Fast algorithm is a kind of algorithm for Corner Detection, and the principle of the algorithm is the test point taken in image, with this
Test point is the center of circle, by the way that the pixel value of the pixel around the center of circle to be compared with the pixel value of test point, to judge
Whether test point is angle point.In embodiments of the present invention, angle point fast algorithm determined is as characteristic point.
S4: being handled characteristic point using sift algorithm, obtains the feature vector of each characteristic point.
In embodiments of the present invention, the quantity of characteristic point can be multiple, wherein feature vector refers to the vector of characteristic point,
It is the amount that characteristic point has size and Orientation, according to the characteristic point that step S3 is obtained, with Scale invariant features transform
(Scale Invariant Feature Transform, sift) algorithm, handles each characteristic point, obtains each spy
Levy the feature vector of point.
Sift algorithm is a kind of Scale invariant features transform algorithm, is mainly used for extracting the locality characteristic of image, pass through
The feature vector that characteristic point is calculated on different scale spaces.
S5: each characteristic point, the legal characteristic point of each of the feature vector of calculating this feature point and legal image are directed to
Feature vector between cosine value, wherein legal image refers to the pre- finger venous image for first passing through legitimate authentication.
In embodiments of the present invention, for each characteristic point, according to the feature vector of characteristic point in gray level image with it is legal
The feature vector of each legal characteristic point in image, the legal spy of each of the feature vector of calculating this feature point and legal image
Levy the cosine value between the feature vector of point.
It should be noted that legal image is stored in advance in valid data library, legal image refer to it is pre- first pass through it is legal
The finger venous image of certification is mainly used to be matched with finger venous image to be detected.
Wherein, legal image characteristic point, legal image characteristic point feature vector can use side identical with this case
Method obtains, and corresponding with legal image is stored in valid data library.
S6: if cosine value is less than preset cosine threshold value, the characteristic point for obtaining the cosine value is confirmed as match point.
In embodiments of the present invention, preset cosine threshold value is mainly used to judge whether match each other between two characteristic points
Point, specific value can be configured according to the needs of practical application, herein with no restrictions.
Specifically, it is compared, when this is remaining by the cosine value being calculated according to step S5 with preset cosine threshold value
When string value is less than preset cosine threshold value, then it is to obtain the legal characteristic point of the cosine value that confirmation, which obtains the characteristic point of the cosine value,
Match point.
S7: if the quantity of match point be greater than preset points threshold value, confirm finger venous image to be detected with it is legal
Image matches.
In embodiments of the present invention, it is preset points threshold value be used for judge two images finger vein grain whether phase
Match, specific value can be configured according to the needs of practical application, herein with no restrictions.
Specifically, the quantity of match point between finger venous image to be detected and legal image is obtained according to step S6,
It is compared with preset points threshold value, when the quantity of the match point is greater than preset points threshold value, is then confirmed to be checked
The finger venous image of survey matches with legal image, i.e., finger venous image to be detected is that homologous finger is quiet with legal image
Arteries and veins image.
It should be noted that by judging whether finger venous image to be detected and legal image are homologous finger vena
Image can be realized the legitimate authentication to finger vena, i.e., when finger venous image to be detected and legal image are homologous hands
When referring to vein image, confirm that the finger vena in the finger venous image to be detected is legal effective finger vena, in turn
It can be the open corresponding permission of user of the finger vena.
In the present embodiment, by carrying out gray processing processing to finger venous image to be detected, gray level image is obtained, is utilized
Fast algorithm extracts the characteristic point of finger vein grain in gray level image, by sift algorithm obtain the feature of each characteristic point to
Amount, and according to feature vector calculates the feature vector and the legal feature of each of legal image of the characteristic point in gray level image
Cosine value between the feature vector of point, the number of match point is judged further according to cosine value, finally by the number of match point and in advance
If threshold value is compared, if the number of match point is greater than preset threshold, finger venous image to be detected and legal figure are confirmed
As matching.On the one hand, the characteristic point that finger vein grain is extracted using fast algorithm, it is real using the high efficiency of fast algorithm
Now to the rapidly extracting of the characteristic point of finger vein grain, to improve the recognition efficiency to finger vein grain characteristic point；Separately
On the one hand, the feature vector of characteristic point is obtained in such a way that fast algorithm and sift algorithm combine, and by calculating feature
The cosine value of vector determines match point, realizes the accurate match to finger vein grain, effectively improves finger vein grain
The accuracy matched.
In one embodiment, as shown in figure 3, in step S2, i.e., finger vein image carries out gray processing processing, obtains ash
Degree image specifically comprises the following steps:
S21: traversing the pixel in finger venous image to be detected, obtains the RGB component of each pixel
Value.
Specifically, the pixel in finger venous image to be detected is traversed according to preset traversal mode, is obtained
Take the RGB component value of each pixel, wherein R, G, B respectively represent the color in three channels of red, green, blue.
Wherein, preset traversal mode specifically can be is with the top left corner pixel point of finger venous image to be detected
Point, sequence from left to right is traversed line by line from top to bottom, is also possible to from the middle line position of finger venous image to be detected
It sets while being traversed to both sides, can also be other traversal modes, herein with no restrictions.
S22: according to the RGB component value of pixel, finger venous image to be detected is made at gray processing according to formula (1)
Reason:
Wherein, x and y is the abscissa and ordinate of each pixel in finger venous image to be detected, and g (x, y) is
Pixel (x, y) gray processing treated gray value, R (x, y) are the color component in the channel R of pixel (x, y), and G (x, y) is
The color component in the channel G of pixel (x, y), B (x, y) are the color component of the channel B of pixel (x, y).
In embodiments of the present invention, in order to realize the accurate extraction to the information content in finger venous image to be detected,
Gray processing processing is carried out firstly the need of to finger venous image to be detected, by formula (1) respectively to the channel R, the channel G and B
The accounting in channel is configured, so that finger venous image to be detected is converted to gray level image.
RGB model is a kind of currently used colour information expression way, it is come using the brightness of Red Green Blue
Quantificational expression color.The model is also referred to as additive color colour mixture model, be with RGB three coloured light mutually the superimposed method to realize colour mixture,
Thus it is suitable for the display of the illuminators such as display.
It should be noted that in embodiments of the present invention, carrying out component by formula (1) and calculating gray value, in other realities
Gray processing processing can also be carried out to image using weighting method, maximum value process or mean value method by applying in example, herein with no restrictions.
In the present embodiment, by traversing the pixel in finger venous image to be detected and obtaining corresponding pixel points
RGB component value, according to the RGB component value of each pixel got, using formula (1) to finger venous image to be detected
Gray processing processing is carried out, the pixel value range of pixel in image is set between 0-255 to realize, is further reduced figure
As original data volume, the computational efficiency in subsequent processing calculating is improved.
In one embodiment, as shown in figure 4, in step S3, i.e., finger vena in gray level image is extracted using fast algorithm
The characteristic point of lines specifically comprises the following steps:
S31: it using the pixel in gray level image as essentially like vegetarian refreshments, and obtains each essentially like the pixel value of vegetarian refreshments.
Specifically, the pixel in gray level image obtained according to preset traversal mode to step S2 traverses, will
The pixel traversed is used as essentially like vegetarian refreshments, also, is obtained each essentially like the pixel value of vegetarian refreshments.
Wherein, preset traversal mode can be sequence from left to right from top to bottom and click through to the pixel in gray level image
Row traverses line by line, is also possible to simultaneously traverse the pixel in gray level image line by line to both sides from intermediate, not do herein
Limitation.
S32: this will be determined as by the pixel on the circumference of radius of pre-set length threshold centered on essentially like vegetarian refreshments
Essentially like the corresponding target pixel points of vegetarian refreshments, and from target pixel points, picture is compared in the target pixel points conduct of selection predeterminated position
Vegetarian refreshments.
In embodiments of the present invention, the object pixel compared with compared pixels point is for clicking through row pixel value with base pixel
Point, by picture circle being carried out by radius of pre-set length threshold, by the circle of the circle centered in step S31 essentially like vegetarian refreshments
All pixels on week, which are selected to be used as, is somebody's turn to do essentially like the corresponding target pixel points of vegetarian refreshments, and the target pixel points for choosing predeterminated position are made
For compared pixels point.
It should be noted that pre-set length threshold can carry out value with the quantity of pixel, specific value can root
It is configured according to the needs of practical application, herein with no restrictions.
It is understood that pre-set length threshold is different, i.e., the radius of circumference is different, the number of the target pixel points on circumference
Amount is also different, for example, 8 target pixel points will be present on circumference if pre-set length threshold is 2 pixels；If default length
Degree threshold value is 3 pixels, then 16 target pixel points will be present on its circumference.
Wherein, predeterminated position can be by two pixels essentially like the vertical direction on the circumference centered on vegetarian refreshments,
It is also possible to can also be by two pixels essentially like the horizontal direction on the circumference centered on vegetarian refreshments essentially like vegetarian refreshments
Centered on circumference on four horizontally and vertically pixel, specific position can be according to practical application
It needs to be configured, herein with no restrictions.
S33: it is corresponding every essentially like vegetarian refreshments with this to be directed to the pixel value calculated each essentially like vegetarian refreshments this essentially like vegetarian refreshments
The first pixel value difference between the pixel value of a compared pixels point, the first pixel value difference of the first preset quantity is greater than pre- if it exists
If first threshold, then this is determined as candidate pixel point essentially like vegetarian refreshments.
It specifically,, will be essentially like vegetarian refreshments according to the compared pixels point that step S32 is obtained for each essentially like vegetarian refreshments
Pixel value with should be essentially like the absolute difference between the pixel value of the corresponding each compared pixels point of vegetarian refreshments as the first pixel difference
Value, then each first pixel value difference is compared with preset first threshold, the first pixel value difference of statistics is greater than first threshold
Quantity, if the quantity be more than or equal to the first preset quantity, confirmation should essentially like vegetarian refreshments be candidate pixel point.
It should be noted that others are all abandoned essentially like vegetarian refreshments, i.e. feature after determining all candidate pixel points
Point need to only be carried out further screening in candidate pixel point and be obtained.
For example, if essentially like vegetarian refreshments Q and comparing picture essentially like vegetarian refreshments Q corresponding four compared pixels point Q1, Q2, Q3 and Q4
The first pixel value difference between vegetarian refreshments Q1, Q2, Q3, Q4 is respectively 8,8,5,8, if preset first threshold is 6, and first pre-
If quantity is 2, then this 3 compared pixels of Q1, Q2 and Q4, which are selected, is greater than preset first with the first pixel value difference essentially like vegetarian refreshments Q
The quantity that threshold value, i.e. the first pixel value difference are greater than first threshold is 3, is greater than the first preset quantity, thus it is confirmed that essentially like
Vegetarian refreshments Q is candidate pixel point.
It should be noted that can preferentially choose during calculating candidate point essentially like the circle centered on vegetarian refreshments
Two compared pixels points of the vertical direction on week are calculated, and filter out first from essentially like vegetarian refreshments according to calculated result
Candidate pixel point, then choose essentially like the four horizontally and vertically compared pixels on the circumference centered on vegetarian refreshments
Point is calculated using identical calculation and the first candidate pixel point, and according to calculated result from the first candidate pixel point
In filter out candidate pixel point again.It is excluded by way of this multiple Stepwise Screening essentially like vegetarian refreshments, can be reduced to time
The calculation amount that pixel is screened is selected, computational efficiency is improved.
It should be noted that the first preset quantity and first threshold can be configured according to the needs of practical application, this
Place is with no restrictions.
S34: it is directed to each candidate pixel point, calculates the pixel value mesh corresponding with the candidate pixel point of the candidate pixel point
The second pixel value difference between the pixel value of pixel is marked, the second pixel value difference of the second preset quantity is greater than the first threshold if it exists
Value, then be determined as characteristic point for the candidate pixel point.
Specifically, for each candidate pixel point, according to the target pixel points that step S32 is obtained, by candidate pixel point
Absolute difference between the pixel value of pixel value each target pixel points corresponding with the candidate pixel point is as the second pixel difference
Value, then each second pixel value difference is compared with preset first threshold, the second pixel value difference of statistics is greater than first threshold
Quantity, if the quantity be more than or equal to the second preset quantity, confirm that the candidate pixel point is characterized a little.
Further, a characteristic point can be only retained in a neighborhood, is existed when in the neighborhood centered on characteristic point
When multiple characteristic points, can calculate separately the corresponding all target pixel points of pixel value of each characteristic point pixel value it
Between absolute difference summation, the maximum characteristic point of summation for choosing absolute difference retained, remaining discarding.Wherein, adjacent
Domain refers to the region that the M*M pixel centered on characteristic point is constituted, and M is the positive integer greater than 1, and the specific value of M can root
It is configured according to the needs of practical application, herein with no restrictions.
In the present embodiment, by obtaining essentially like vegetarian refreshments, obtain essentially like the corresponding target pixel points of vegetarian refreshments, and obtain
Compared pixels point, it is corresponding essentially like vegetarian refreshments with this essentially like the pixel value of vegetarian refreshments by calculating for each essentially like vegetarian refreshments
Pixel value difference between the pixel value of each compared pixels point determines candidate pixel point, then is directed to each candidate pixel point, calculates
The second pixel value difference between the pixel value of the pixel value of candidate pixel point target pixel points corresponding with the candidate pixel point, from
And determine characteristic point.By first determining candidate pixel point, can effective exclusive segment non-characteristic point, recycle candidate pixel point
The pixel value of the corresponding target pixel points of pixel value is compared, and can reduce calculation amount, so as to improve to feature
The computational efficiency of point realizes the rapidly extracting to characteristic point.
In one embodiment, as shown in figure 5, in step S4, i.e., characteristic point is handled using sift algorithm, is obtained every
The feature vector of a characteristic point specifically comprises the following steps:
S41: the sampling area of each characteristic point is obtained.
Specifically, each characteristic point got according to step S3 chooses B*B specification centered on each characteristic point
Sampling area of the region as each characteristic point, wherein specification identifies the size of image-region, and B is the number of pixel, and
Its specific value can be configured according to the needs of practical application, herein with no restrictions.
S42: sampling area is rotated according to preset direction, obtains object region, and by object region
It is divided into the subregion of predetermined number.
Specifically, sampling area is rotated according to preset direction according to formula (2):
Wherein, x and y is respectively the original x-axis and y-axis of sampling area, and α is the direction of rotation, and x' and y' are respectively to sample
Region is according to the postrotational x' axis of preset direction and y' axis.
The image-region obtained after sampling area is rotated according to formula (2) is determined as object region, and right
The object region is divided according to equally spaced mode, i.e., the region of B*B specification is divided into predetermined number at equal intervals
Subregion, that is, be divided into BP*BPSub-regions.
It should be noted that preset direction can be specifically configured according to the needs of practical application, herein with no restrictions.
Predetermined number is the positive integer greater than 1.
S43: the gradient orientation histogram feature of each subregion is calculated, wherein gradient orientation histogram feature includes n
Gradient magnitude on preset direction, n are positive integer.
Specifically, the gradient direction and gradient magnitude of pixel in each subregion are calculated according to formula (3):
Wherein, x and y is the abscissa and ordinate of each pixel in each subregion, and V (x, y) is pixel (x, y)
Gradient magnitude in the horizontal direction, H (x, y) are the gradient magnitude of pixel (x, y) in vertical direction, and β (x, y) is pixel
The gradient direction of point (x, y), m (x, y) are the gradient magnitude of pixel (x, y).
In each subregion, the gradient direction of each pixel and gradient width in the subregion are obtained according to formula (3)
Value, the gradient direction of each pixel is mapped on some corresponding preset direction according to preset direction scope.For example,
If the gradient direction of pixel E is 30 °, the gradient direction of pixel F is 80 °, it is assumed that preset direction includes 45 ° and 90 °, and
Be mapped to direction scope on 45 ° of preset directions be (22.5 °, 67.5 °], the direction scope being mapped on 90 ° of preset directions is
(67.5 °, 112.5 °], then pixel E is mapped on 45 ° of preset direction, pixel F is mapped to 90 ° of preset direction
On.
After the gradient direction of each pixel is mapped on corresponding preset direction, by pixel each in the subregion
The gradient magnitude of the same preset direction of point adds up, as the gradient intensity information of the preset direction, i.e. gradient magnitude, and with
For the starting point of all preset directions as being combined with a starting point, the gradient orientation histogram for obtaining each subregion is special
Sign, which includes the gradient intensity information on n preset direction and each preset direction.
For example, it is assumed that n preset direction includes 0 °, 90 °, 180 ° and 270 ° totally 4 preset direction, if depositing in a subregion
In five pixels Q1, Q2, Q3, Q4 and Q5, corresponding preset direction is respectively 0 °, 0 °, 90 °, 180 ° and 270 °, corresponding ladder
Spending amplitude is respectively 1,2,1,1 and 1, then the gradient magnitude of same preset direction adds up, i.e., by pixel Q1 and pixel
The gradient magnitude of point Q2 adds up, then after the starting point of all preset directions is combined as same a starting point, finally
To obtain one has 0 °, 90 °, 180 ° and 270 ° this four preset directions, and the gradient magnitude of each preset direction is respectively
3,1,1,1 plane vector, the plane vector are gradient orientation histogram feature.
S44: the gradient orientation histogram feature of each subregion is combined, obtain the foundation characteristic of characteristic point to
Amount.
Specifically, the gradient orientation histogram feature in each subregion is combined, since each subregion includes n
The gradient intensity information in a direction, and there are Bp*BpSub-regions, by the way that B will be formed after combinationP*BP* n dimension feature to
Amount, this feature vector is the foundation characteristic vector of characteristic point.
For example, if the sampling area of characteristic point there are 4 sub-regions, and in each subregion all exist 8 directions ladder
Strength information is spent, by will obtain the feature vector of 32 dimensions after combination, i.e. the feature vector of 32 dimension is the base of this feature point
Plinth feature vector.
S45: foundation characteristic vector is normalized, feature vector is obtained.
In embodiments of the present invention, normalized mode can be each foundation characteristic vector is special divided by all bases
The maximum value in vector is levied, it can also be by each foundation characteristic vector divided by the mean value of all foundation characteristic vectors.By to base
Plinth feature vector is normalized, and obtains the feature vector of standard, convenient to handle data.
Further, after normalized, truncation can also be carried out to the length of feature vector, if feature vector
Length be greater than preset length, then take preset length as the length of feature vector, only so as to improve the distinctive of data processing.
Wherein, preset length can be configured according to the needs of practical application, herein with no restrictions.
For example, if preset length is 0.2, the length of feature vector is 0.5, by carrying out truncation to this feature vector
Afterwards, the length of this feature vector is 0.2.
In the present embodiment, rotated to obtain object region by the sampling area to each characteristic point, further
Object region is divided into the subregion of predetermined number, and calculates the gradient orientation histogram feature of each subregion, then
The foundation characteristic vector that characteristic point is obtained after the gradient orientation histogram feature of each subregion is combined, finally to basis
Feature vector is normalized, and obtains feature vector.On the one hand, by the way that object region is divided into multiple sub-districts
Domain, and the gradient magnitude of the different gradient directions of calculating to each subregion, can accurately extract the feature of subregion, and then right
The foundation characteristic vector that the gradient orientation histogram feature of each subregion obtains after being combined being capable of accurate response feature
The feature of point improves the precision of characteristic point feature vector so as to analyze according to the foundation characteristic vector characteristic point；
On the other hand, by the way that foundation characteristic vector is normalized, it can guarantee the distinctive of data processing, to realize to feature
The accurate extraction of the feature vector of point.
In one embodiment, in step S5, that is, each characteristic point, the feature vector of calculating this feature point and legal figure are directed to
Cosine value between each of the picture feature vector of legal characteristic point specifically comprises the following steps:
The feature vector of each characteristic point and each legal feature in legal image in gray level image are calculated according to formula (4)
Cosine value between the feature vector of point:
Wherein, cos θijFor j-th of legal feature in the feature vector of ith feature point in gray level image and legal image
Cosine value between the feature vector of point, aiFor the feature vector of ith feature point in gray level image, bjFor jth in legal image
The feature vector of a legal characteristic point, | ai| it is aiLength, | bj| it is bjLength, i and j are positive integer.
Specifically, according to each legal characteristic point in the feature vector of characteristic point each in gray level image and legal image
Feature vector calculates the feature vector of each characteristic point and each conjunction in legal image in finger venous image according to formula (4)
Cosine value between the feature vector of method characteristic point.
In the present embodiment, using formula (4) can rapidly and accurately calculate the feature vector of each characteristic point with it is legal
Cosine value in image between the feature vector of each legal characteristic point utilizes the cosine value accurate judgement characteristic point and legal spy
Feature similarity degree between sign point, and then match point is accurately identified, the matched accuracy of finger vein grain is improved, thus
When finger vein image carries out legal conscientious, the accuracy of certification is effectively improved.
In one embodiment, as shown in fig. 6, after step S6, which further includes following steps:
S80: if the quantity of match point is less than or equal to points threshold value, it is corresponding with legal image to calculate match point
Relative displacement mean value between legal characteristic point, wherein relative displacement mean value includes the relatively legal image of gray level image in horizontal seat
The x-component mean value that mark side is offset up, and the upper y-component mean value deviated in the ordinate.
In embodiments of the present invention, when the quantity of match point is less than or equal to points threshold value, in order to avoid due to figure
As quality problems etc. lead to that it fails to match, finger venous image and legal figure further can be judged by relative displacement again
The matching degree of picture.
Specifically, the opposite position between match point and legal characteristic point corresponding in legal image is calculated according to formula (5)
Move mean value:
Wherein, x and y is the abscissa of each characteristic point and vertical in each legal characteristic point and gray level image in legal image
Coordinate, Pe(x, y) is opposite between e-th of match point legal characteristic point point corresponding with legal image in gray level image
Displacement, P2e(x, y) is the position of e-th of characteristic point in gray level image, P1e(x, y) is the position of e-th of characteristic point in legal image
It sets, relative displacement mean value of the bias between all match points and legal characteristic point corresponding in legal image, o is gray level image
The sum of middle match point, wherein the value range of e is [1, o].
The phase between match point and legal characteristic point corresponding in legal image in gray level image is obtained according to formula (5)
To displacement mean value, the overlapping region of the relatively legal image of gray level image is obtained on abscissa direction by the relative displacement mean value
The x-component mean value of offset, and the upper y-component mean value deviated in the ordinate.
S81: according to x-component mean value and y-component mean value, the overlapping region between gray level image and legal image is determined.
In embodiments of the present invention, the overlay region by the available legal image of x-component mean value relative to gray level image
The distance that domain is offset up in abscissa side, the overlapping by the available legal image of y-component mean value relative to gray level image
The region upper distance deviated, and then the overlapping region between acquisition gray level image and legal image in the ordinate.
For example, width 10 is cross with length direction using the gray level image lower left corner as origin if a length of the 20 of gray level image
Coordinate direction, then the length of gray level image is abscissa 0 to 20, using width direction as ordinate direction, then the width in gray scale direction
Degree is ordinate 0 to 10, and the overlapping region of image versus grayscale image offsets by the right 10 on abscissa direction if legal,
5 have been offset up in ordinate direction, then overlapping region on abscissa direction between abscissa 10 to 20, in ordinate side
Upwards between ordinate 5 to 10.
S82: in overlapping region, according to gray level image and legal image between the pixel value of same position pixel
Pixel value difference calculates the number for mismatching point.
Specifically, obtain the overlapping region between gray level image and legal image according to step S81, using gray level image with
Pixel value difference of the legal image between the pixel value of same position pixel calculates the number for mismatching point according to formula (6):
Wherein, NmTo mismatch number a little in overlapping region, x and y are pixel in legal image and gray level image
Abscissa and ordinate, and legal image and gray level image be all in the same coordinate system, the lower left corner pixel and ash of legal image
The lower left corner pixel of image is spent in the coordinate origin of coordinate system, and s is the overlapping region of legal image versus grayscale image
The mean value of the x-component of offset on the horizontal scale, t are overlapping region on the vertical scale inclined of legal image versus grayscale image
The mean value of the y-component of shifting, h are the height of gray level image and legal image, and w is the width of gray level image and legal image, P1And P2
It is all the pixel value of pixel, I (x, y) is the pixel value of pixel (x, y) in gray level image, and R (x-s, y-t) is legal image
The pixel value of middle pixel (x-s, y-t).
When gray level image with legal image when the absolute value of the pixel value of same position pixel differs 255, i.e., one
For white pixel point, one is black pixel point, then the two pixels are unmatched pixel, using formula (6), is calculated
Show that gray level image and legal image mismatch number a little in overlapping region.
S83: according to the number of the pixel of presetted pixel value in the number and overlapping region for mismatching point, grayscale image is calculated
Picture and the mismatch degree between legal image.
In embodiments of the present invention, the pixel of presetted pixel value is specifically as follows gray level image pixel in overlapping region
The pixel and the legal image pixel that pixel value is 255 in overlapping region that value is 255.
Specifically, gray level image is obtained according to step S82 and legal image mismatches number a little in overlapping region, with
And in overlapping region the pixel of presetted pixel value number, calculated between gray level image and legal image according to formula (7)
Mismatch degree:
Wherein, RmFor the mismatch degree of gray level image and legal image.
S84: if mismatch degree is less than preset matching degree threshold value, confirm finger venous image to be detected and legal figure
As matching.
Specifically, the mismatch degree that gray level image and legal image are obtained according to step S83, by itself and preset matching degree
Threshold value is compared, if the mismatch degree be less than preset matching degree threshold value, confirm finger venous image to be detected with
Legal image matches, i.e., finger venous image to be detected and legal image are homologous finger venous images.
It should be noted that the preset specific value of matching degree threshold value can be set according to the needs of practical application
It sets, herein with no restrictions.
In the present embodiment, by calculating in gray level image between match point and legal characteristic point corresponding in legal image
Relative displacement mean value obtains the x-component mean value that the overlapping region of legal image versus grayscale image is offset up in abscissa side,
And the y-component mean value above deviated in the ordinate, so that it is determined that the overlapping region between gray level image and legal image, and
Gray level image is calculated in overlapping region and legal image mismatches the number of point, then the number by mismatching point calculates gray scale
Mismatch degree is finally compared by the mismatch degree between image and legal image with preset matching degree threshold value, if not
It is less than preset matching degree threshold value with degree, then confirms that finger venous image to be detected matches with legal image.When match point
When quantity is less than or equal to points threshold value, is further matched, avoided due to image again in such a way that displacement calculates
Quality problems lead to that it fails to match, so as to effectively improve to the matched accuracy of finger vein grain, and then pass through judgement
Whether finger venous image to be detected matches with legal image, can be realized the legitimate authentication to finger vena.
It should be understood that the size of the serial number of each step is not meant that the order of the execution order in above-described embodiment, each process
Execution sequence should be determined by its function and internal logic, the implementation process without coping with the embodiment of the present invention constitutes any limit
It is fixed.
In one embodiment, a kind of image matching apparatus is provided, image in the image matching apparatus and above-described embodiment
Method of completing the square corresponds.As shown in fig. 7, the image matching apparatus includes: to obtain module 71, gray scale module 72, feature point extraction
Module 73, vector calculation module 74, similarity calculation module 75, identification module 76 and the first matching module 77.Each functional module
Detailed description are as follows:
Module 71 is obtained, for obtaining finger venous image to be detected；
Gray scale module 72 obtains gray level image for carrying out gray processing processing to finger venous image to be detected；
Feature point extraction module 73, for using fast algorithm to extract the characteristic point of finger vein grain in gray level image；
Vector calculation module 74 obtains the feature of each characteristic point for handling using sift algorithm characteristic point
Vector；
Similarity calculation module 75, for being directed to each characteristic point, the feature vector of calculating this feature point and legal image
Each of cosine value between the feature vector of legal characteristic point, wherein legal image refers to the pre- legitimate authentication that first passes through
Finger venous image；
Identification module 76, it is if being less than preset cosine threshold value for cosine value, the characteristic point for obtaining the cosine value is true
Think match point；
First matching module 77 confirms hand to be detected if the quantity for match point is greater than preset points threshold value
Refer to that vein image matches with legal image.
Further, gray scale module 72 includes:
It traverses submodule 721 and obtains each picture for traversing to the pixel in finger venous image to be detected
The RGB component value of vegetarian refreshments；
Submodule 722 is handled, it is quiet to finger to be detected according to following formula for the RGB component value according to pixel
Arteries and veins image makees gray processing processing:
Wherein, x and y is the abscissa and ordinate of each pixel in finger venous image to be detected, and g (x, y) is
Pixel, (x, y) are gray processing treated gray value, and R (x, y) is the color component in the channel R of pixel (x, y), G (x,
Y) for pixel (x, y) the channel G color component, B (x, y) be pixel (x, y) channel B color component.
Further, feature point extraction module 73 includes:
Set submodule 731, for being used as the pixel in gray level image essentially like vegetarian refreshments, and acquisition each essentially like
The pixel value of vegetarian refreshments；
Submodule 732 is chosen, being used for will be centered on essentially like vegetarian refreshments, using pre-set length threshold as on the circumference of radius
Pixel is determined as this essentially like the corresponding target pixel points of vegetarian refreshments, and the target picture of predeterminated position is chosen from target pixel points
Vegetarian refreshments is as compared pixels point；
Candidate sub-block 733, it is basic with this for being directed to the pixel value calculated each essentially like vegetarian refreshments this essentially like vegetarian refreshments
The first pixel value difference between the pixel value of the corresponding each compared pixels point of pixel, if it exists the first preset quantity first
Pixel value difference is greater than preset first threshold, then this is determined as candidate pixel point essentially like vegetarian refreshments；
Feature submodule 734 calculates pixel value and the candidate of the candidate pixel point for being directed to each candidate pixel point
The second pixel value difference between the pixel value of the corresponding target pixel points of pixel, if it exists the second preset quantity the second pixel
Difference is greater than first threshold, then the candidate pixel point is determined as characteristic point.
Further, vector calculation module 74 includes:
Submodule 741 is sampled, for obtaining the sampling area of each characteristic point；
Submodule 742 is rotated, for rotating sampling area according to preset direction, obtains object region, and
Object region is divided into the subregion of predetermined number；
Histogram calculation submodule 743, for calculating the gradient orientation histogram feature of each subregion, wherein gradient
Direction histogram feature includes the gradient magnitude on n preset direction, and n is positive integer；
It combines submodule 744 and obtains characteristic point for the gradient orientation histogram feature of each subregion to be combined
Foundation characteristic vector；
Normalizing submodule 745 obtains feature vector for foundation characteristic vector to be normalized.
Further, similarity calculation module 75 includes:
Cosine computational submodule 751, for calculating the feature vector of each characteristic point in gray level image according to following formula
Cosine value between the feature vector of legal characteristic point each in legal image:
Wherein, cos θijFor j-th of legal feature in the feature vector of ith feature point in gray level image and legal image
Cosine value between the feature vector of point, aiFor the feature vector of ith feature point in gray level image, bjFor jth in legal image
The feature vector of a legal characteristic point, | ai| it is aiLength, | bj| it is bjLength, i and j are positive integer.
Further, the image matching apparatus further include:
Be displaced computing module 78, if for match point quantity be less than or equal to points threshold value, calculate match point with
Relative displacement mean value in legal image between corresponding legal characteristic point, wherein relative displacement mean value includes gray level image phase
To the x-component mean value that legal image is offset up in abscissa side, and the upper y-component mean value deviated in the ordinate；
Area determination module 79, for according to x-component mean value and y-component mean value, determine gray level image and legal image it
Between overlapping region；
Pixel difference computing module 710 is used in overlapping region, according to gray level image and legal image in same position picture
Pixel value difference between the pixel value of vegetarian refreshments calculates the number for mismatching point；
Mismatch degree computing module 711, for the picture according to presetted pixel value in the number and overlapping region for mismatching point
The number of vegetarian refreshments calculates the mismatch degree between gray level image and legal image；
Second matching module 712 confirms finger to be detected if being less than preset matching degree threshold value for mismatch degree
Vein image matches with legal image.
Specific about image matching apparatus limits the restriction that may refer to above for image matching method, herein not
It repeats again.Modules in above-mentioned image matching apparatus can be realized fully or partially through software, hardware and combinations thereof.On
Stating each module can be embedded in the form of hardware or independently of in the processor in computer equipment, can also store in a software form
In memory in computer equipment, the corresponding operation of the above modules is executed in order to which processor calls.
In one embodiment, a kind of computer equipment is provided, which can be server, internal junction
Composition can be as shown in Figure 8.The computer equipment include by system bus connect processor, memory, network interface and
Database.Wherein, the processor of the computer equipment is for providing calculating and control ability.The memory packet of the computer equipment
Include non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system, computer program and data
Library.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating
The database of machine equipment is used to store the data of finger venous image.The network interface of the computer equipment is used for and external end
End passes through network connection communication.To realize a kind of image matching method when the computer program is executed by processor.
In one embodiment, a kind of computer equipment is provided, including memory, processor and storage are on a memory
And the computer program that can be run on a processor, processor realize above-described embodiment images match side when executing computer program
The step of method, such as step S1 shown in Fig. 2 to step S7.Alternatively, processor realizes above-mentioned implementation when executing computer program
The function of each module/unit of image matching apparatus in example, such as module 71 shown in Fig. 7 is to the function of module 77.To avoid weight
Multiple, which is not described herein again.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Image matching method in above method embodiment is realized when machine program is executed by processor, alternatively, the computer program is processed
The function of each module in image matching apparatus in above-mentioned apparatus embodiment is realized when device executes.It is no longer superfluous here to avoid repeating
It states.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
To any reference of memory, storage, database or other media used in each embodiment provided herein,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
It is apparent to those skilled in the art that for convenience of description and succinctly, only with above-mentioned each function
Can unit, module division progress for example, in practical application, can according to need and by above-mentioned function distribution by different
Functional unit, module are completed, i.e., the internal structure of described device is divided into different functional unit or module, more than completing
The all or part of function of description.
Embodiment described above is merely illustrative of the technical solution of the present invention, rather than its limitations；Although referring to aforementioned reality
Applying example, invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each
Technical solution documented by embodiment is modified or equivalent replacement of some of the technical features；And these are modified
Or replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution should all
It is included within protection scope of the present invention.
Claims (10)
1. a kind of image matching method, which is characterized in that described image matching process includes:
Obtain finger venous image to be detected；
Gray processing processing is carried out to the finger venous image to be detected, obtains gray level image；
The characteristic point of finger vein grain in the gray level image is extracted using fast algorithm；
The characteristic point is handled using sift algorithm, obtains the feature vector of each characteristic point；
For each characteristic point, the feature vector and each of legal image legal characteristic point of the characteristic point are calculated
Cosine value between feature vector, wherein the legal image refers to the pre- finger venous image for first passing through legitimate authentication；
If the cosine value is less than preset cosine threshold value, the characteristic point for obtaining the cosine value is confirmed as match point；
If the quantity of the match point be greater than preset points threshold value, confirm the finger venous image to be detected with it is described
Legal image matches.
2. image matching method as described in claim 1, which is characterized in that described to the finger venous image to be detected
Gray processing processing is carried out, obtaining gray level image includes:
Pixel in the finger venous image to be detected is traversed, the RGB component of each pixel is obtained
Value；
According to the RGB component value of the pixel, gray processing is made to the finger venous image to be detected according to following formula
Processing:
Wherein, x and y be the finger venous image to be detected in each pixel abscissa and ordinate, g (x,
It y) is pixel (x, the y) gray processing treated gray value, R (x, y) is the color in the channel R of the pixel (x, y)
Component, G (x, y) are the color component in the channel G of the pixel (x, y), and B (x, y) is the channel B of the pixel (x, y)
Color component.
3. image matching method as described in claim 1, which is characterized in that described to extract the grayscale image using fast algorithm
The characteristic point of finger vein grain includes: as in
Using the pixel in the gray level image as essentially like vegetarian refreshments, and obtain each pixel value essentially like vegetarian refreshments；
Will by it is described essentially like vegetarian refreshments centered on, be determined as the base by the pixel on the circumference of radius of pre-set length threshold
The corresponding target pixel points of this pixel, and the target pixel points conduct of selection predeterminated position is compared from the target pixel points
Pixel；
For each described essentially like vegetarian refreshments, calculate the pixel value essentially like vegetarian refreshments with it is described corresponding often essentially like vegetarian refreshments
The first pixel value difference between the pixel value of a compared pixels point, if it exists the first preset quantity first pixel difference
Value is greater than preset first threshold, then is determined as candidate pixel point essentially like vegetarian refreshments for described；
For each candidate pixel point, the pixel value institute corresponding with the candidate pixel point of the candidate pixel point is calculated
The second pixel value difference between the pixel value of target pixel points is stated, the second preset quantity second pixel value difference is big if it exists
In the first threshold, then the candidate pixel point is determined as the characteristic point.
4. image matching method as described in claim 1, which is characterized in that described to be clicked through using sift algorithm to the feature
Row processing, the feature vector for obtaining each characteristic point include:
Obtain the sampling area of each characteristic point；
The sampling area is rotated according to preset direction, obtains object region, and by the object region
It is divided into the subregion of predetermined number；
Calculate the gradient orientation histogram feature of each subregion, wherein the gradient orientation histogram feature includes n
Gradient magnitude on preset direction, n are positive integer；
The gradient orientation histogram feature of each subregion is combined, the foundation characteristic of the characteristic point is obtained
Vector；
The foundation characteristic vector is normalized, described eigenvector is obtained.
5. image matching method as described in claim 1, which is characterized in that it is described to be directed to each characteristic point, calculate institute
The cosine value stated between the feature vector of characteristic point and each of the legal image feature vector of legal characteristic point includes:
It is calculated in the gray level image according to following formula every in the feature vector of each characteristic point and the legal image
Cosine value between the feature vector of a legal characteristic point:
Wherein, cos θijFor the feature vector of ith feature point in the gray level image and j-th in the legal image it is legal
The cosine value between the feature vector of characteristic point, aiFor the feature vector of i-th of characteristic point in the gray level image,
bjFor the feature vector of j-th of legal characteristic point in the legal image, | ai| it is aiLength, | bj| it is bjLength, i
It is positive integer with j.
6. the image matching method as described in any one of claim 1 to 5, which is characterized in that if the cosine value
Less than preset cosine threshold value, then after the characteristic point for obtaining the cosine value being confirmed as match point, described image matching process
Further include:
If the quantity of the match point is less than or equal to the points threshold value, the match point and the legal image are calculated
In relative displacement mean value between corresponding legal characteristic point, wherein the relative displacement mean value includes the gray level image phase
To the x-component mean value that the legal image is offset up in abscissa side, and in the ordinate, the upper y-component deviated is equal
Value；
According to the x-component mean value and the y-component mean value, determine overlapping between the gray level image and the legal image
Region；
In the overlapping region, according to the gray level image and the legal image same position pixel pixel value it
Between pixel value difference, calculate mismatch point number；
According to the number of the pixel of presetted pixel value in the number for mismatching point and the overlapping region, the ash is calculated
Spend the mismatch degree between image and the legal image；
If the mismatch degree is less than preset matching degree threshold value, the finger venous image to be detected and the conjunction are confirmed
Method image matches.
7. a kind of image matching apparatus, which is characterized in that described image coalignment includes:
Module is obtained, for obtaining finger venous image to be detected；
Gray scale module obtains gray level image for carrying out gray processing processing to the finger venous image to be detected；
Feature point extraction module, for extracting the characteristic point of finger vein grain in the gray level image using fast algorithm；
Vector calculation module obtains the spy of each characteristic point for handling using sift algorithm the characteristic point
Levy vector；
Similarity calculation module, for be directed to each characteristic point, calculate the characteristic point feature vector and legal image
Each of cosine value between the feature vector of legal characteristic point, wherein the legal image refers to that pre- first pass through legal is recognized
The finger venous image of card；
Identification module confirms the characteristic point for obtaining the cosine value if being less than preset cosine threshold value for the cosine value
For match point；
First matching module confirms described to be detected if the quantity for the match point is greater than preset points threshold value
Finger venous image matches with the legal image.
8. image matching apparatus as claimed in claim 7, which is characterized in that the computing module includes:
Cosine computational submodule, for calculating the feature vector of each characteristic point in the gray level image according to following formula
Cosine value between the feature vector of the legal characteristic point each in the legal image:
Wherein, cos θijFor the feature vector of ith feature point in the gray level image and j-th in the legal image it is legal
The cosine value between the feature vector of characteristic point, aiFor the feature vector of i-th of characteristic point in the gray level image,
bjFor the feature vector of j-th of legal characteristic point in the legal image, | ai| it is aiLength, | bj| it is bjLength, i
It is positive integer with j.
9. a kind of computer equipment, including memory, processor and storage are in the memory and can be in the processor
The computer program of upper operation, which is characterized in that the processor realized when executing the computer program as claim 1 to
The step of any one of 5 described image matching process.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, and feature exists
In the step of any one of such as claim 1 to 5 of realization described image matching process when the computer program is executed by processor
Suddenly.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810593681.5A CN109101867A (en) | 2018-06-11 | 2018-06-11 | A kind of image matching method, device, computer equipment and storage medium |
| PCT/CN2018/104050 WO2019237520A1 (en) | 2018-06-11 | 2018-09-05 | Image matching method and apparatus, computer device, and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810593681.5A CN109101867A (en) | 2018-06-11 | 2018-06-11 | A kind of image matching method, device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109101867A true CN109101867A (en) | 2018-12-28 |
Family
ID=64796799
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810593681.5A Withdrawn CN109101867A (en) | 2018-06-11 | 2018-06-11 | A kind of image matching method, device, computer equipment and storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN109101867A (en) |
| WO (1) | WO2019237520A1 (en) |
Cited By (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109977909A (en) * | 2019-04-04 | 2019-07-05 | 山东财经大学 | Finger vein identification method and system based on minutiae point Region Matching |
| CN110070564A (en) * | 2019-05-08 | 2019-07-30 | 广州市百果园信息技术有限公司 | A kind of characteristic point matching method, device, equipment and storage medium |
| CN110162953A (en) * | 2019-05-31 | 2019-08-23 | Oppo（重庆）智能科技有限公司 | Biometric discrimination method and Related product |
| CN110162954A (en) * | 2019-05-31 | 2019-08-23 | Oppo（重庆）智能科技有限公司 | Right management method and Related product |
| CN110188815A (en) * | 2019-05-24 | 2019-08-30 | 广州市百果园信息技术有限公司 | A kind of characteristic point method of sampling, device, equipment and storage medium |
| CN110210394A (en) * | 2019-05-31 | 2019-09-06 | Oppo广东移动通信有限公司 | Vein image acquisition method and Related product |
| CN110287161A (en) * | 2019-07-02 | 2019-09-27 | 北京字节跳动网络技术有限公司 | Image processing method and device |
| CN110310399A (en) * | 2019-06-26 | 2019-10-08 | Oppo广东移动通信有限公司 | The unlocking method and Related product of vein door lock |
| CN110348289A (en) * | 2019-05-27 | 2019-10-18 | 广州中国科学院先进技术研究所 | A kind of finger vein identification method based on binary map |
| CN110633733A (en) * | 2019-08-14 | 2019-12-31 | 中国平安财产保险股份有限公司 | Intelligent image matching method and device and computer readable storage medium |
| CN110689485A (en) * | 2019-10-14 | 2020-01-14 | 中国空气动力研究与发展中心超高速空气动力研究所 | SIFT image splicing method applied to infrared nondestructive testing of large pressure container |
| CN110738236A (en) * | 2019-09-16 | 2020-01-31 | 深圳市国信合成科技有限公司 | Image matching method and device, computer equipment and storage medium |
| CN110766095A (en) * | 2019-11-01 | 2020-02-07 | 易思维（杭州）科技有限公司 | Defect detection method based on image gray level features |
| CN110851349A (en) * | 2019-10-10 | 2020-02-28 | 重庆金融资产交易所有限责任公司 | Page abnormal display detection method, terminal equipment and storage medium |
| CN110866503A (en) * | 2019-11-19 | 2020-03-06 | 圣点世纪科技股份有限公司 | Abnormality detection method and system for finger vein equipment |
| CN110880003A (en) * | 2019-10-12 | 2020-03-13 | 中国第一汽车股份有限公司 | Image matching method and device, storage medium and automobile |
| CN112270372A (en) * | 2020-11-06 | 2021-01-26 | 首都师范大学 | Method, device, computer equipment and medium for determining target object |
| CN113048912A (en) * | 2021-02-26 | 2021-06-29 | 山东师范大学 | Calibration system and method for projector |
| CN113469971A (en) * | 2021-06-30 | 2021-10-01 | 深圳中科飞测科技股份有限公司 | Image matching method, image matching detection device and storage medium |
| CN115063613A (en) * | 2022-08-09 | 2022-09-16 | 海纳云物联科技有限公司 | Method and device for verifying commodity label |
| CN115331212A (en) * | 2022-10-13 | 2022-11-11 | 南通东鼎彩印包装厂 | Method for identifying abnormal code spraying of zip-top can bottom |
| CN115953407A (en) * | 2023-03-15 | 2023-04-11 | 深圳市科姆特精密科技有限公司 | Semiconductor equipment maintenance system based on computer vision |
| CN117037159A (en) * | 2023-10-09 | 2023-11-10 | 网思科技股份有限公司 | Oil painting authenticity identification method, device and storage medium based on convolutional neural network |
| CN117893502A (en) * | 2024-01-15 | 2024-04-16 | 广州市科帕电子科技有限公司 | Image detection method, device, equipment and storage medium |
| CN117930375A (en) * | 2024-03-22 | 2024-04-26 | 国擎(山东)信息科技有限公司 | Multi-dimensional detection technology fused channel type terahertz human body security inspection system |
Families Citing this family (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111223134B (en) * | 2020-01-19 | 2023-04-14 | 北京航空航天大学 | Feature vector construction method of linear features, feature matching method and computing equipment |
| CN111445402B (en) * | 2020-03-20 | 2023-09-19 | 上海中通吉网络技术有限公司 | Image denoising method and device |
| CN111462074B (en) * | 2020-03-31 | 2024-03-19 | 珠海格力智能装备有限公司 | Bearing appearance detection method, device, system, computer equipment and storage medium |
| CN113658261B (en) * | 2020-05-12 | 2024-07-30 | 武汉Tcl集团工业研究院有限公司 | Checkerboard corner position detection method and device, intelligent terminal and storage medium |
| CN111695498B (en) * | 2020-06-10 | 2023-04-07 | 西南林业大学 | Wood identity detection method |
| CN111815578B (en) * | 2020-06-23 | 2024-05-28 | 浙江大华技术股份有限公司 | Image stripe detection method, video monitoring system and related device |
| CN112084883B (en) * | 2020-08-18 | 2024-08-13 | 珠海格力电器股份有限公司 | Finger vein collection method, device, finger vein identification equipment and storage medium |
| CN112164032B (en) * | 2020-09-14 | 2023-12-29 | 浙江华睿科技股份有限公司 | Dispensing method and device, electronic equipment and storage medium |
| CN112150522A (en) * | 2020-09-22 | 2020-12-29 | 上海商汤智能科技有限公司 | Remote sensing image registration method, device, equipment, storage medium and system |
| CN112308886A (en) * | 2020-09-27 | 2021-02-02 | 深圳疆程技术有限公司 | Method, device and system for determining HUD image size |
| CN112381105B (en) * | 2020-11-16 | 2024-05-07 | 平安科技（深圳）有限公司 | Automatic processing method, device, equipment and medium for slide block verification |
| CN112686302B (en) * | 2020-12-29 | 2024-02-13 | 科大讯飞股份有限公司 | Image feature point matching method, computer equipment and storage device |
| CN112883827B (en) * | 2021-01-28 | 2024-03-29 | 腾讯科技（深圳）有限公司 | Method and device for identifying specified target in image, electronic equipment and storage medium |
| CN112950540B (en) * | 2021-02-01 | 2022-04-15 | 联宝(合肥)电子科技有限公司 | Bar code identification method and equipment |
| CN113033550B (en) * | 2021-03-15 | 2022-04-15 | 合肥联宝信息技术有限公司 | Image detection method and device and computer readable medium |
| CN113592921B (en) * | 2021-06-30 | 2024-06-21 | 北京旷视科技有限公司 | Image matching method, device, electronic equipment and storage medium |
| CN113627446B (en) * | 2021-08-18 | 2023-10-31 | 成都工业学院 | Image matching method and system of feature point description operator based on gradient vector |
| CN114217756B (en) * | 2021-11-26 | 2023-09-19 | 海信电子科技(武汉)有限公司 | Display device, image detection method and device |
| CN115314634B (en) * | 2022-06-28 | 2024-05-31 | 上海艾为电子技术股份有限公司 | Anti-shake detection method, device, terminal equipment and readable storage medium |
| CN115184380B (en) * | 2022-08-01 | 2024-04-26 | 国网上海市电力公司 | Method for detecting abnormity of welding spots of printed circuit board based on machine vision |
| CN115830719B (en) * | 2023-02-16 | 2023-04-28 | 青岛旭华建设集团有限公司 | Building site dangerous behavior identification method based on image processing |
| CN116310448B (en) * | 2023-05-24 | 2023-08-04 | 山东曙岳车辆有限公司 | Container assembly matching detection method based on computer vision |
| CN116452586B (en) * | 2023-06-15 | 2023-09-26 | 山东飞宏工程机械有限公司 | Automatic butt welding quality detection system for tunnel small guide pipe residual materials |
| CN116612306B (en) * | 2023-07-17 | 2023-09-26 | 山东顺发重工有限公司 | Computer vision-based intelligent flange plate alignment method and system |
| CN116703909B (en) * | 2023-08-07 | 2023-10-27 | 威海海泰电子有限公司 | Intelligent detection method for production quality of power adapter |
| CN117292142B (en) * | 2023-09-26 | 2024-05-10 | 北京鹰之眼智能健康科技有限公司 | Image matching method |
| CN116993643B (en) * | 2023-09-27 | 2023-12-12 | 山东建筑大学 | Unmanned aerial vehicle photogrammetry image correction method based on artificial intelligence |
| CN117495852B (en) * | 2023-12-29 | 2024-05-28 | 天津中荣印刷科技有限公司 | Digital printing quality detection method based on image analysis |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101470800B (en) * | 2007-12-30 | 2011-05-04 | 沈阳工业大学 | Hand shape recognition method |
| CN103268483B (en) * | 2013-05-31 | 2017-08-04 | 沈阳工业大学 | Palm grain identification method under open environment non-contact capture |
| CN106529407A (en) * | 2016-10-09 | 2017-03-22 | 北京电子科技职业学院 | Vehicle-mounted fingerprint recognition method |
| CN107273844A (en) * | 2017-06-12 | 2017-10-20 | 成都芯软科技股份公司 | Vena metacarpea recognizes matching process and device |
-
2018
- 2018-06-11 CN CN201810593681.5A patent/CN109101867A/en not_active Withdrawn
- 2018-09-05 WO PCT/CN2018/104050 patent/WO2019237520A1/en active Application Filing
Cited By (39)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109977909A (en) * | 2019-04-04 | 2019-07-05 | 山东财经大学 | Finger vein identification method and system based on minutiae point Region Matching |
| CN110070564A (en) * | 2019-05-08 | 2019-07-30 | 广州市百果园信息技术有限公司 | A kind of characteristic point matching method, device, equipment and storage medium |
| CN110070564B (en) * | 2019-05-08 | 2021-05-11 | 广州市百果园信息技术有限公司 | Feature point matching method, device, equipment and storage medium |
| CN110188815A (en) * | 2019-05-24 | 2019-08-30 | 广州市百果园信息技术有限公司 | A kind of characteristic point method of sampling, device, equipment and storage medium |
| CN110348289A (en) * | 2019-05-27 | 2019-10-18 | 广州中国科学院先进技术研究所 | A kind of finger vein identification method based on binary map |
| CN110162954A (en) * | 2019-05-31 | 2019-08-23 | Oppo（重庆）智能科技有限公司 | Right management method and Related product |
| CN110210394A (en) * | 2019-05-31 | 2019-09-06 | Oppo广东移动通信有限公司 | Vein image acquisition method and Related product |
| CN110210394B (en) * | 2019-05-31 | 2021-05-04 | Oppo广东移动通信有限公司 | Vein image acquisition method and related product |
| CN110162954B (en) * | 2019-05-31 | 2021-09-28 | Oppo（重庆）智能科技有限公司 | Authority management method and related product |
| CN110162953A (en) * | 2019-05-31 | 2019-08-23 | Oppo（重庆）智能科技有限公司 | Biometric discrimination method and Related product |
| CN110310399A (en) * | 2019-06-26 | 2019-10-08 | Oppo广东移动通信有限公司 | The unlocking method and Related product of vein door lock |
| CN110310399B (en) * | 2019-06-26 | 2021-11-02 | Oppo广东移动通信有限公司 | Unlocking method of vein door lock and related product |
| CN110287161A (en) * | 2019-07-02 | 2019-09-27 | 北京字节跳动网络技术有限公司 | Image processing method and device |
| CN110633733A (en) * | 2019-08-14 | 2019-12-31 | 中国平安财产保险股份有限公司 | Intelligent image matching method and device and computer readable storage medium |
| CN110633733B (en) * | 2019-08-14 | 2024-05-03 | 中国平安财产保险股份有限公司 | Image intelligent matching method, device and computer readable storage medium |
| CN110738236A (en) * | 2019-09-16 | 2020-01-31 | 深圳市国信合成科技有限公司 | Image matching method and device, computer equipment and storage medium |
| CN110738236B (en) * | 2019-09-16 | 2022-07-22 | 深圳市国信合成科技有限公司 | Image matching method and device, computer equipment and storage medium |
| CN110851349A (en) * | 2019-10-10 | 2020-02-28 | 重庆金融资产交易所有限责任公司 | Page abnormal display detection method, terminal equipment and storage medium |
| CN110880003A (en) * | 2019-10-12 | 2020-03-13 | 中国第一汽车股份有限公司 | Image matching method and device, storage medium and automobile |
| CN110689485A (en) * | 2019-10-14 | 2020-01-14 | 中国空气动力研究与发展中心超高速空气动力研究所 | SIFT image splicing method applied to infrared nondestructive testing of large pressure container |
| CN110766095A (en) * | 2019-11-01 | 2020-02-07 | 易思维（杭州）科技有限公司 | Defect detection method based on image gray level features |
| CN110766095B (en) * | 2019-11-01 | 2022-04-19 | 易思维（杭州）科技有限公司 | Defect detection method based on image gray level features |
| CN110866503A (en) * | 2019-11-19 | 2020-03-06 | 圣点世纪科技股份有限公司 | Abnormality detection method and system for finger vein equipment |
| CN110866503B (en) * | 2019-11-19 | 2024-01-05 | 圣点世纪科技股份有限公司 | Abnormality detection method and abnormality detection system for finger vein equipment |
| CN112270372A (en) * | 2020-11-06 | 2021-01-26 | 首都师范大学 | Method, device, computer equipment and medium for determining target object |
| CN112270372B (en) * | 2020-11-06 | 2023-09-29 | 首都师范大学 | Method, device, computer equipment and medium for determining target object |
| CN113048912A (en) * | 2021-02-26 | 2021-06-29 | 山东师范大学 | Calibration system and method for projector |
| CN113469971A (en) * | 2021-06-30 | 2021-10-01 | 深圳中科飞测科技股份有限公司 | Image matching method, image matching detection device and storage medium |
| CN113469971B (en) * | 2021-06-30 | 2023-10-13 | 深圳中科飞测科技股份有限公司 | Image matching method, detection device and storage medium |
| CN115063613A (en) * | 2022-08-09 | 2022-09-16 | 海纳云物联科技有限公司 | Method and device for verifying commodity label |
| CN115331212A (en) * | 2022-10-13 | 2022-11-11 | 南通东鼎彩印包装厂 | Method for identifying abnormal code spraying of zip-top can bottom |
| CN115331212B (en) * | 2022-10-13 | 2023-10-27 | 南通东鼎彩印包装厂 | Method for identifying abnormal spraying code at bottom of pop can |
| CN115953407B (en) * | 2023-03-15 | 2023-06-09 | 深圳市科姆特精密科技有限公司 | Semiconductor equipment maintenance system based on computer vision |
| CN115953407A (en) * | 2023-03-15 | 2023-04-11 | 深圳市科姆特精密科技有限公司 | Semiconductor equipment maintenance system based on computer vision |
| CN117037159A (en) * | 2023-10-09 | 2023-11-10 | 网思科技股份有限公司 | Oil painting authenticity identification method, device and storage medium based on convolutional neural network |
| CN117037159B (en) * | 2023-10-09 | 2024-03-19 | 网思科技股份有限公司 | Oil painting authenticity identification method, device and storage medium based on convolutional neural network |
| CN117893502A (en) * | 2024-01-15 | 2024-04-16 | 广州市科帕电子科技有限公司 | Image detection method, device, equipment and storage medium |
| CN117930375A (en) * | 2024-03-22 | 2024-04-26 | 国擎(山东)信息科技有限公司 | Multi-dimensional detection technology fused channel type terahertz human body security inspection system |
| CN117930375B (en) * | 2024-03-22 | 2024-06-25 | 国擎(山东)信息科技有限公司 | Multi-dimensional detection technology fused channel type terahertz human body security inspection system |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2019237520A1 (en) | 2019-12-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109101867A (en) | A kind of image matching method, device, computer equipment and storage medium | |
| Ye et al. | Robust registration of multimodal remote sensing images based on structural similarity | |
| Woo et al. | Improving color constancy in an ambient light environment using the phong reflection model | |
| US8036468B2 (en) | Invariant visual scene and object recognition | |
| JP4868530B2 (en) | Image recognition device | |
| CN110546651B (en) | Method, system and computer readable medium for identifying objects | |
| CN108876723B (en) | Method for constructing color background of gray target image | |
| CN107945111B (en) | Image stitching method based on SURF (speeded up robust features) feature extraction and CS-LBP (local binary Pattern) descriptor | |
| Oliveira et al. | A probabilistic approach for color correction in image mosaicking applications | |
| CN108133212A (en) | A kind of quota invoice amount identifying system based on deep learning | |
| CN110427746A (en) | Sliding block verifies code verification method, device, storage medium and computer equipment | |
| AU2020203790B2 (en) | Transformed multi-source content aware fill | |
| Oliveira Santos et al. | Automatic mapping of cracking patterns on concrete surfaces with biological stains using hyper‐spectral images processing | |
| Ghannam et al. | Cross correlation versus mutual information for image mosaicing | |
| US20040202385A1 (en) | Image retrieval | |
| CN111192194A (en) | Panoramic image splicing method for curtain wall building vertical face | |
| CN111582155B (en) | Living body detection method, living body detection device, computer equipment and storage medium | |
| Peng et al. | Hyperspectral image stitching via optimal seamline detection | |
| CN109801281A (en) | A kind of texture-free metal parts image regulation straightway edge detection method | |
| Markiewicz | The use of computer vision algorithms for automatic orientation of terrestrial laser scanning data | |
| Cai et al. | An adaptive symmetry detection algorithm based on local features | |
| CN107145820B (en) | Binocular positioning method based on HOG characteristics and FAST algorithm | |
| CN109961393A (en) | Subpixel registration and splicing based on interpolation and iteration optimization algorithms | |
| Jasvilis et al. | Finding complex patterns using template matching | |
| Hurtík et al. | Fuzzy transform theory in the view of image registration application |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20181228 |
|
| WW01 | Invention patent application withdrawn after publication |