CN107563293A - A kind of new finger vena preprocess method and system - Google Patents
A kind of new finger vena preprocess method and system Download PDFInfo
- Publication number
- CN107563293A CN107563293A CN201710656229.4A CN201710656229A CN107563293A CN 107563293 A CN107563293 A CN 107563293A CN 201710656229 A CN201710656229 A CN 201710656229A CN 107563293 A CN107563293 A CN 107563293A
- Authority
- CN
- China
- Prior art keywords
- image
- processing
- extraction
- matrix
- finger
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of new finger vena preprocess method and system, can obtain de-redundancy and protrude the finger vein image data of venous information.This method includes carrying out finger areas extraction process to original image, obtains image after region of interesting extraction；Image enhancement processing is carried out to image after the region of interesting extraction using genetic algorithm, obtains enhancing image；Data Dimensionality Reduction is carried out to the enhancing image to handle with de-redundancy, obtain dimensionality reduction and the matrix image after whitening processing using principal component analysis algorithm and Zero phase Component Analysis albefactions algorithms.By the embodiment of the present invention, carry out extracted region, image enhaucament, view data dimensionality reduction and de-redundancy successively to original image and handle, the finger vein image data of signature can be obtained.
Description
Technical field
The present invention relates to technical field of image processing, more particularly to a kind of new finger vena preprocess method and it is
System.
Background technology
Living things feature recognition is to utilize physical characteristics collecting device and computer technology, the physiology based on human body inherently
Feature or behavioural characteristic carry out the means of personal identification discriminating.Wherein finger vein identification technology belongs to the life based on physiological characteristic
Important a member in thing identification.Medical research is proved everyone finger vein grain and differed, and same person is not
Also differed with finger.High safety to refer to vein promise well with live body.National standard appraisal meeting is vessel graph within 2014
As data with formally including living things feature recognition data interchange format 9 parts.Referring to the development of vein technology to China has great meaning
Justice.
One completely refers to vein recognition system often comprising IMAQ, image preprocessing, feature extraction, characteristic matching four
Part.Wherein image preprocessing is an important ring, and the height of pretreating effect quality directly determines experimental result below.
Peking University's Huang, Lee et al. propose a kind of finger hand vein recognition scheme based on curve detector, and wherein preprocessing part uses
Image normalization, reduce picture size and cause arithmetic speed faster, then carry out feature extraction.But their pretreatment is only
It is to carry out simple image normalization, gives no thought to the foreground zone and background area of collection image, can only be applicable and image quality
The fine and image not comprising noise.Japanese Scientists Higashi et al. proposes a kind of spy for being based on curve tracing
Levy extracting method and the application in hand vein recognition is referred to, wherein pretreatment portion, which is divided into, enters line translation to the position of image and angle and make
It is equal to obtain position and the angle of all input pictures, curve tracing directly then is carried out to input picture.Preprocessing part pair
Picture position and angle enter line translation so that all input pictures have equal position and angle, afterwards by simple normalization
Feature extraction is directly carried out later, directly have ignored influence caused by noise, under different pieces of information collection, the feature extracted includes
A large amount of non-finger venous informations, and influenceed seriously by light, contrast.
Image preprocessing in existing finger vein identification technology is mostly filtered, the segmentation of image preprocessing, image
Deng operand is huge and often fails to extract venous information or feature well.For realizing what contrast strengthened by filtering,
Operand is huge and often fails to protrude venous information for more complicated image.In existing finger vein pretreating scheme,
Generally be normalized to referring to vein image, binaryzation, filtering, the scheme such as refinement, preprocessing process complexity and can not be preferable
Prominent venous information, lacks universality, is generally only operated under specific database, is showed if collecting device is changed very poor.
The content of the invention
It is an object of the invention to overcome the deficiencies in the prior art, and the invention provides a kind of new finger vena to locate in advance
Method and system are managed, the finger vein image of signature can be obtained.
In order to solve the above-mentioned technical problem, a kind of new finger vena preprocess method of the embodiment of the present invention, the side
Method includes：
Finger areas extraction process is carried out to original image, obtains image after region of interesting extraction；
Using genetic algorithm to carrying out image enhancement processing after the region of interesting extraction, enhancing image is obtained；
Using principal component analysis algorithm and Zero-phase Component Analysis albefactions algorithms to the enhancing
Image carries out Data Dimensionality Reduction and handled with de-redundancy, obtains dimensionality reduction and the matrix image after whitening processing.
Preferably, it is described that original image progress finger areas extraction process is comprised the following steps：
1) gray processing processing is carried out to the original image using rgb2gray, obtains two-dimensional matrix image after gray processing,
Wherein rgb2gray formula are as follows：
Gray=0.3R+0.59G+0.11B
Wherein Gray represents grey, and R represents red, and G represents green, and B represents blueness；
2) Edge extraction processing is carried out to two-dimensional matrix image after gray processing using sobel edge detection operators, obtained
Take 0,1 bianry image；
3) contour extraction processing is carried out to described 0,1 bianry image using two values matrix contours extract scheme, obtains profile
0,1 bianry image after extraction；
4) branch process is removed to 0,1 bianry image after the contours extract, obtains image after removal branch；
5) finger train of thought extension processing is carried out to the incomplete finger edge of image after the removal branch, obtains train of thought and extend
Image afterwards, image is image after region of interesting extraction after the train of thought extends.
Preferably, it is described that described 0,1 bianry image progress extraction process is included using two values matrix contours extract scheme：
Described 0,1 bianry image is retrieved since the image upper left corner, runs into the value for 1, starts to detect whether to be continuous
Curve, if full curve then records coordinate in Matrix C, so as to be obtained by the fraction area of absence filled in discontinuous curve
To full curve, i.e. finger edge profile, obtain profile matrix C, C comprising two road wheel profiles and represent as follows：
C=[C (1) C (2) ... C (k) ... C (N)]
The form of wherein each contour line represents as follows：
C (k)=[level1 x1 x2 x3 ... numxy y1 y2 y3 ...]
Wherein level1 is brightness value grade, and numxy is profile point number, and each (x, y) determines a bit on profile
(x, y), therefore obtain 0,1 bianry image after contours extract.
Preferably, it is described branch process is removed to described 0,1 bianry image to include：
To the horizontal direction of described 0,1 bianry image as base, along described 0,1 bianry image vertical direction set up one it is perpendicular
Straight detection line, described 0,1 bianry image from left to right records the intersection point number for referring to vein edge image and vertical detector
CountA and the number in the range of three pixels of vertical line or so, if number is more than presetting threshold value, vertical detection
Number is set as 0 entirely in the range of the pixel and the pixel of left and right three that device passes through；
To described 0,1 bianry image vertical direction as base, level detection is set up along described 0,1 bianry image horizontal direction
Line, described 0,1 bianry image contains pixel in the range of recording level detection three pixels of line and horizontal line or so from top to bottom
Point value is 1 number, is determined as edge if number exceedes presetting threshold value, is otherwise branch, pixel point value is complete in branch
Portion is set to 0.
Preferably, the incomplete finger edge of the image to after the removal branch, which carries out the extension of finger train of thought, includes：
To the image after the removal branch since middle part with detecting vertical curve, run into discrete point and then judge
Pixel number in horizontal line adjacent domain, if pixel number exceedes pre-set threshold, with the horizontal position of surrounding pixel point
Put and coordinate of the average of vertical position as current discontinuity point；
If four corners of the image after the removal branch are free of continuity point, record refers to the angle of inclination at vein edge
It is as follows if the upper right corner, calculation formula for θ：
Y (f (i+tan (θ/360*2* π) * (k-j)), k)=1
Wherein, y represents image array, and f represents bracket function, and (i, j) represents pixel, and k represents to wait the coordinate for extending point.
Preferably, it is described that image progress image enhancement processing after the region of interesting extraction is included using genetic algorithm
Following steps：
1) chromosome structure is initialized, produces one group of n random number, for the scope of random number between 0 to 255, n represents dye
The size of colour solid, the number of gray value grade in the extracted region images is represented, to random number sequence and first element
0 is set to, nth elements are set to 255；
2) corresponding picture number of individuals is randomly generated, as the first generation, the chromosome element of the first generation is obtained by the first step；
3) edge of the extracted region images is calculated using fitness function, obtains the adaptation number of degrees of each chromosome
Value, according to ranking fitness individual；
4) present age is selected and bred, mating is included by wheel disc method assortative mating individual, breeding according to fitness
With mutation, produce individual of future generation and form colony, wherein what is do not mated directly arrives the next generation；
5) the 4) step is repeated, until meeting that end condition reaches iterations, records the chromosome structure finally given；
6) chromosome structure is mapped using tonal gradation mapping, obtains contrast enhancing image, the contrast
Enhancing image is input picture.
Preferably, it is described to be calculated using principal component analysis algorithm and Zero-phase Component Analysis albefactions
Method carries out Data Dimensionality Reduction to the enhancing image and comprised the following steps with de-redundancy processing：
1) the enhancing picture size is normalized to 40*80, then cutting is 10 × 10 fritter, and sliding window is each
Mobile 5 units, and each fritter is preserved into vector；
2) each vector is averaged, then each element subtracts average in vector, and the vector of acquisition is designated as x；
3) sigma value is obtained, calculation formula is as follows：
4) the eigenvectors matrix U and eigenvalue matrix S of ∑ are calculated using singular value decomposition, each row of matrix U include one
Individual characteristic vector, S are diagonal matrix, the characteristic vector of same number of columns in each value homography U on its diagonal, vector
XPCAwhite is that the vectorial calculation formula after dimensionality reduction is as follows：
XPCAwhite=diag (1./sqrt (diag (S)+epsilon)) * U'*x
Wherein, epsilon is constant, is set as 3*10-5；
5) vector x ZCAwhite is calculated, i.e. dimensionality reduction and the vector after whitening processing, calculation formula is as follows：
XZCAwhite=U*xPCAwhite
To the further solution of formula：
XZCAwhite=U*diag (1./sqrt (diag (S)+epsilon)) * U'*x
Wherein, U is the characteristic vector obtained by step 4), and S is the matrix that diagonal includes characteristic value, and x is input number
According to by obtaining xZCAwhite, a sub-picture is transformed to the matrix image of a secondary size reduction.
In addition, the embodiment of the present invention additionally provides a kind of new finger vena pretreatment system, the system includes：
Area image extraction module：For carrying out finger areas extraction process to original image, obtain area-of-interest and carry
Take rear image；
Image enhancement module：For image after the region of interesting extraction to be carried out into image enhancement processing, enhancing is obtained
Image；
Image dimensionality reduction and whitening processing module：Handle, obtain with de-redundancy for the enhancing image to be carried out into Data Dimensionality Reduction
Take dimensionality reduction and the matrix image after whitening processing.
Preferably, the area image extraction mould includes：
Image gray processing processing unit：For the original image to be carried out into gray processing processing, two dimension after gray processing is obtained
Matrix image；
Edge extraction processing unit：For two-dimensional matrix image after the gray processing to be carried out at Edge extraction
Reason, obtain 0,1 bianry image；
Image outline extraction process unit：For described 0,1 bianry image to be carried out into contour extraction processing, obtain profile and carry
0,1 bianry image after taking；
Image debranching enzyme processing unit：For 0,1 bianry image after the contours extract to be removed into branch process, obtain
Remove except image after branch；
Finger train of thought extends processing unit：For the incomplete finger edge of image after the removal branch to be carried out into finger arteries and veins
Network extension is handled, and obtains image after train of thought extends.
Preferably, described image enhancing module includes：
Chromosome initialization unit：For producing one group of n random number, between 0 to 255, n is represented the scope of random number
The size of chromosome, the number of gray value grade in the extracted region images is represented, to random number sequence and first member
Element is set to 0, and nth elements are set to 255；
Generation chromosome acquiring unit：For randomly generating corresponding picture number of individuals, as the first generation, the dye of the first generation
Colour solid element is obtained by the first step；
Chromosome fitness computing unit：For calculating the edge of the extracted region images, each chromosome is obtained
Fitness numerical value, according to ranking fitness individual；
Individual choice and diaspore：For being selected the present age and being bred, selected according to fitness by wheel disc method
Mating individual, breeding include mating and mutation, produce individual of future generation and form colony, wherein what is do not mated directly arrives down
A generation；
Chromosome structure map unit：For mapping chromosome structure, contrast enhancing image is obtained.
By carrying out extracted region successively to original image, image enhaucament, view data dimensionality reduction and de-redundancy are handled, can be with
Obtain the finger vein image of signature.Wherein extracted region ensures to obtain region interested under different acquisition device, i.e.,
Only comprising the finger areas for referring to venous information, and be normalized to identical size, contrast enhancement process ensure to highlight refer to it is quiet
Arteries and veins information, carrying out dimensionality reduction and albefaction to image reduces data dimension, reduces the expense of feature extraction and matching.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
There is the required accompanying drawing used in technology description to be briefly described, it is clear that, drawings in the following description are only this
Some embodiments of invention, for those of ordinary skill in the art, on the premise of not paying creative work, can be with
Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 is a kind of schematic flow sheet of new finger vena preprocess method of the embodiment of the present invention；
The S11 that Fig. 2 is Fig. 1 carries out finger areas extraction process to original image, obtains image after region of interesting extraction
Detailed process schematic diagram；
The S12 that Fig. 3 is Fig. 1 carries out image enhancement processing using genetic algorithm to image after the region of interesting extraction,
Obtain the detailed process schematic diagram of enhancing image；
Fig. 4 is a kind of structural representation of new finger vena pretreatment system of the embodiment of the present invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained all other under the premise of creative work is not made
Embodiment, belong to the scope of protection of the invention.
Fig. 1 is a kind of schematic flow sheet of new finger vena preprocess method of the embodiment of the present invention, as shown in figure 1,
Methods described includes：
S11：Finger areas extraction process is carried out to original image, obtains image after region of interesting extraction；
S12：Image enhancement processing is carried out to image after the region of interesting extraction using genetic algorithm, obtains enhancing figure
Picture；
S13：Using principal component analysis algorithm and Zero-phase Component Analysis albefaction algorithms to described
Strengthen image and carry out Data Dimensionality Reduction and de-redundancy processing, obtain dimensionality reduction and the matrix image after whitening processing.
S11 is further illustrated：
To original image carry out finger areas extraction process, obtain region of interesting extraction after image detailed process by scheming
Shown in 2.
S12 is further illustrated：
Image enhancement processing is carried out to image after the region of interesting extraction using genetic algorithm, obtains enhancing image
Detailed process is as shown in Figure 3.
S13 is further illustrated：
The correlation of adjacent elements may impact to feature extraction, and some features extracted may be useless, in order to
Reduce data redundancy and adjacent element correlation, using principal component analysis (Principal Component Analysis,
) and ZCA albefaction algorithms, PCA comprise the following steps that:
1) the enhancing picture size is normalized to 40*80, then cutting is 10 × 10 fritter, and sliding window is each
Mobile 5 units, and each fritter is preserved into vector；
2) each vector is averaged, then each element subtracts average in vector, and the vector of acquisition is designated as x；
3) sigma value is obtained, calculation formula is as follows：
4) after by 3) obtaining sigma, the eigenvectors matrix U and eigenvalue matrix S of ∑ are calculated using singular value decomposition,
The each row of matrix U include a characteristic vector, and S is diagonal matrix, same number of columns in each value homography U on its diagonal
Characteristic vector, vector x PCAwhite is that the vectorial calculation formula after dimensionality reduction is as follows：
XPCAwhite=diag (1./sqrt (diag (S)+epsilon)) * U'*x
Wherein, epsilon is constant, is set as 3*10-5；
5) vector x ZCAwhite is calculated, i.e. dimensionality reduction and the vector after whitening processing, calculation formula is as follows：
XZCAwhite=U*xPCAwhite
To the further solution of formula：
XZCAwhite=U*diag (1./sqrt (diag (S)+epsilon)) * U'*x
Wherein, U is the characteristic vector obtained by step 4), and S is the matrix that diagonal includes characteristic value, and x is input number
According to the matrix obtained by obtaining xZCAwhite as to be carried out to x after dimensionality reduction and de-redundancy.By this step, a sub-picture becomes
The matrix image of a secondary size reduction is changed to, feature extraction and matching step can be entered.
The S11 that Fig. 2 is Fig. 1 carries out finger areas extraction process to original image, obtains image after region of interesting extraction
Detailed process schematic diagram, as shown in Fig. 2 methods described includes：
S111：Gray processing processing is carried out to the original image using rgb2gray gray processings, obtains two dimension after gray processing
Matrix image；
S112：Edge extraction processing is carried out to two-dimensional matrix image after gray processing using sobel edge detection operators,
Obtain 0,1 bianry image；
S113：Contour extraction processing is carried out to described 0,1 bianry image using two values matrix contours extract scheme, obtains wheel
0,1 bianry image after exterior feature extraction；
S114：Branch process is removed to 0,1 bianry image after the contours extract, obtains image after removal branch；
S115：Finger train of thought extension processing is carried out to the incomplete finger edge of image after the removal branch, obtains train of thought
Image after extension, image is image after region of interesting extraction after the train of thought extends；
S111 is further illustrated：
Gray processing processing is carried out to the original image using rgb2gray, obtains two-dimensional matrix image after gray processing, its
Middle rgb2gray formula are as follows：
Gray=0.3R+0.59G+0.11B
Wherein Gray represents grey, and R represents red, and G represents green, and B represents blueness.
S112 is further illustrated：
Edge extraction processing is carried out to two-dimensional matrix image after gray processing using sobel edge detection operators, obtained
0,1 bianry image.
S113 is further illustrated：
In real life, everyone finger edge is continuous profile, in the absence of the situation of interruption, based on this visitor
See true.Bianry image after the 2nd step edge extracting is retrieved since the image upper left corner, runs into the value for 1, starts to detect
Whether it is full curve, if full curve then records coordinate in Matrix C.In addition influenceed, gathered by illumination, collection posture
Obtained profile is generally incomplete imperfect, is obtained continuously by the fraction area of absence filled in discontinuous curve in the present invention
Curve, i.e. finger edge profile, obtain containing the continuous segment in image in the profile matrix C, C comprising two road wheel profiles, such as
There is n continuous segment in fruit image, then n full curve is calculated, Matrix C represents as follows：
C=[C (1) C (2) ... C (k) ... C (N)]
The form of wherein each contour line represents as follows：
C (k)=[level1 x1 x2 x3 ... numxy y1 y2 y3 ...]
Wherein level1 is brightness value grade, and numxy is profile point number, and each (x, y) determines a bit on profile
(x, y), therefore 0,1 bianry image after contours extract is obtained, two values matrix is recorded as, the value note of the pixel on edge contour
For 1,0 is otherwise designated as.
S114 is further illustrated：
According to the ductility and flatness of finger contours, remove the discontinuous and inappropriate branch of angle and do not meet hand
Refer to the branch in extension direction, specific method includes：
To the horizontal direction of described 0,1 bianry image as base, along described 0,1 bianry image vertical direction set up one it is perpendicular
Straight detection line, described 0,1 bianry image from left to right records the intersection point number for referring to vein edge image and vertical detector
CountA and the number in the range of three pixels of vertical line or so, if number is more than presetting threshold value, vertical detection
Number is set as 0 entirely in the range of the pixel and the pixel of left and right three that device passes through；
To described 0,1 bianry image vertical direction as base, level detection is set up along described 0,1 bianry image horizontal direction
Line, described 0,1 bianry image contains pixel in the range of recording level detection three pixels of line and horizontal line or so from top to bottom
Point value is 1 number, is determined as edge if number exceedes presetting threshold value, is otherwise branch, pixel point value is complete in branch
Portion is set to 0.
S115 is further illustrated：
Due to the image that collects, illumination condition is unstable sometimes, and the same finger of same person is in different collections
Moment putting position, angle etc. may be different, it is difficult to be extracted completely by rim detection and contours extract from image
Finger contours, thus also need to incompleteness finger edge according to finger other parts angular characteristicses carry out finger lengthening.
Specific method is as follows：
To the image after the removal branch since middle part with detecting vertical curve, run into discrete point and then judge
Pixel number in horizontal line adjacent domain, if pixel number exceedes pre-set threshold, with the horizontal position of surrounding pixel point
Put and coordinate of the average of vertical position as current discontinuity point；
If four corners of the image after the removal branch are free of continuity point, record refers to the angle of inclination at vein edge
It is as follows if the upper right corner, calculation formula for θ：
Y (f (i+tan (θ/360*2* π) * (k-j)), k)=1
Wherein, y represents image array, and f represents bracket function, and (i, j) represents pixel, and k represents to wait the coordinate for extending point.
The S12 that Fig. 3 is Fig. 1 carries out image enhancement processing using genetic algorithm to image after the region of interesting extraction,
The detailed process schematic diagram of enhancing image is obtained, as shown in figure 3, methods described includes：
S121：Chromosome structure is initialized, produces one group of n random number, the scope of random number is between 0 to 255；
S122：Corresponding picture number of individuals is randomly generated, as the first generation；
S123：The edge of the extracted region images is calculated using fitness function, obtains the fitness of each chromosome
Numerical value, according to ranking fitness individual；
S124：The present age is selected and bred, according to fitness by wheel disc method assortative mating individual, is produced of future generation
Individual simultaneously forms colony, wherein what is do not mated directly arrives the next generation.Repeat, until meeting end condition, record is most
The chromosome structure obtained eventually；
S125：Chromosome structure is mapped using tonal gradation mapping, obtains contrast enhancing image, the contrast
Degree enhancing image is input picture.
S121 is further illustrated：
Initial chromosome structure defines:One group of n random number is produced, for the scope of random number between 0 to 255, n is dyeing
The size of body, the number of gray value grade in input picture is represented, 0, n-th is set to random number sequence and first element
Individual element is set to 255.
S122 is further illustrated：
Corresponding picture number of individuals is randomly generated, as the first generation, the element numerical value of first generation chromosome is obtained by the first step
Arrive.
S123 is further illustrated：
Fitness numerical value is calculated by fitness function, according to ranking fitness individual, fitness calculation formula is such as
Under：
Fitness (x)=log (log (E (I (X)))) * n_edges (I (X))
Wherein, fitness (x) represents adaptive value, and I (X) is enhancing image, and n_edges (I (X)) represents original process
SOBEL rim detection detectors detect enhancing image number of edges, E (I (X)) be enhancing image intensity value and,
Log-log operations are used for avoiding producing unnatural image.
S124 is further illustrated：
The present age is selected and bred, individual can be all evaluated in every generation.Select to hand over by wheel disc method according to fitness
With individual, breeding includes mating and mutation, produces individual of future generation and forms colony, wherein directly arriving for not mated is next
Generation.
This step is constantly repeated, until meeting that end condition reaches iterations, the chromosome structure that will finally give
Record.
S125 is further illustrated：
Chromosome structure is mapped using tonal gradation mapping, grayscale mapping function is as follows：
T (G (k))=Ci(k) k=1,2,3 ..., n
Wherein T represents that G is the gray value sequence of input picture, and k is represented for changing the function of original image gray value
The number of input gray level value grade.C is a complete chromosome, Ci(k) value of k-th gene is represented.Grayscale mapping function
For a sub-picture is remapped, enhanced image is obtained.
Fig. 4 is a kind of structural representation of new finger vena pretreatment system of the embodiment of the present invention, as shown in figure 4,
The system includes：
11：Area image extraction module, for carrying out finger areas extraction process to original image, obtain area-of-interest
Image after extraction；
12：Image enhancement module, for image after the region of interesting extraction to be carried out into image enhancement processing, obtain and increase
Strong image；
13：Image dimensionality reduction and whitening processing module, handled for the enhancing image to be carried out into Data Dimensionality Reduction with de-redundancy,
Obtain dimensionality reduction and the matrix image after whitening processing.
Further illustrated to 11：
The area image extraction mould includes：
Image gray processing processing unit：For the original image to be carried out into gray processing processing, two dimension after gray processing is obtained
Matrix image；
Edge extraction processing unit：For two-dimensional matrix image after the gray processing to be carried out at Edge extraction
Reason, obtain 0,1 bianry image；
Image outline extraction process unit：For described 0,1 bianry image to be carried out into contour extraction processing, obtain profile and carry
0,1 bianry image after taking；
Image debranching enzyme processing unit：For 0,1 bianry image after the contours extract to be removed into branch process, obtain
Remove except image after branch；
Finger train of thought extends processing unit：For the incomplete finger edge of image after the removal branch to be carried out into finger arteries and veins
Network extension is handled, and obtains image after train of thought extends.
Further illustrated to 12：
Described image enhancing module includes：
Chromosome initialization unit：For producing one group of n random number, between 0 to 255, n is represented the scope of random number
The size of chromosome, the number of gray value grade in the extracted region images is represented, to random number sequence and first member
Element is set to 0, and nth elements are set to 255；
Generation chromosome acquiring unit：For randomly generating corresponding picture number of individuals, as the first generation, the dye of the first generation
Colour solid element is obtained by the first step；
Chromosome fitness computing unit：For calculating the edge of the extracted region images, each chromosome is obtained
Fitness numerical value, according to ranking fitness individual；
Individual choice and diaspore：For being selected the present age and being bred, selected according to fitness by wheel disc method
Mating individual, breeding include mating and mutation, produce individual of future generation and form colony, wherein what is do not mated directly arrives down
A generation；
Chromosome structure map unit：For mapping chromosome structure, contrast enhancing image is obtained.
Further illustrated to 13：
Using principal component analysis (Principal Component Analysis, PCA) and ZCA albefaction algorithms, specifically
Step is as follows:
1) the enhancing picture size is normalized to 40*80, then cutting is 10 × 10 fritter, and sliding window is each
Mobile 5 units, and each fritter is preserved into vector；
2) each vector is averaged, then each element subtracts average in vector, and the vector of acquisition is designated as x；
3) sigma value is obtained, calculation formula is as follows：
4) after by 3) obtaining sigma, the eigenvectors matrix U and eigenvalue matrix S of ∑ are calculated using singular value decomposition,
The each row of matrix U include a characteristic vector, and S is diagonal matrix, same number of columns in each value homography U on its diagonal
Characteristic vector, vector x PCAwhite is that the vectorial calculation formula after dimensionality reduction is as follows：
XPCAwhite=diag (1./sqrt (diag (S)+epsilon)) * U'*x
Wherein, epsilon is constant, is set as 3*10-5；
5) vector x ZCAwhite is calculated, i.e. dimensionality reduction and the vector after whitening processing, calculation formula is as follows：
XZCAwhite=U*xPCAwhite
To the further solution of formula：
XZCAwhite=U*diag (1./sqrt (diag (S)+epsilon)) * U'*x
Wherein, U is the characteristic vector obtained by step 4), and S is the matrix that diagonal includes characteristic value, and x is input number
According to the matrix obtained by obtaining xZCAwhite as to be carried out to x after dimensionality reduction and de-redundancy.
Specifically, the operation principle of the system related functions module of the embodiment of the present invention can be found in the correlation of embodiment of the method
Description, is repeated no more here.
By carrying out extracted region successively to original image, image enhaucament, view data dimensionality reduction and de-redundancy are handled, can be with
Obtain the finger vein image of signature.Wherein extracted region ensures to obtain region interested under different acquisition device, i.e.,
Only comprising the finger areas for referring to venous information, and be normalized to identical size, contrast enhancement process ensure to highlight refer to it is quiet
Arteries and veins information, carrying out dimensionality reduction and albefaction to image reduces data dimension, reduces the expense of feature extraction and matching.
One of ordinary skill in the art will appreciate that all or part of step in the various methods of above-described embodiment is can
To instruct the hardware of correlation to complete by program, the program can be stored in a computer-readable recording medium, storage
Medium can include：Read-only storage (ROM, Read Only Memory), random access memory (RAM, Random
Access Memory), disk or CD etc..
In addition, a kind of new finger vena preprocess method and system that are provided above the embodiment of the present invention are carried out
It is discussed in detail, herein should employs specific case and the principle and embodiment of the present invention are set forth, the above is implemented
The explanation of example is only intended to help the method and its core concept for understanding the present invention；Meanwhile for the general technology people of this area
Member, according to the thought of the present invention, there will be changes in specific embodiments and applications, in summary, this explanation
Book content should not be construed as limiting the invention.
Claims (10)
1. a kind of new finger vena preprocess method, it is characterised in that methods described includes：
Finger areas extraction process is carried out to original image, obtains image after region of interesting extraction；
Image enhancement processing is carried out to image after the region of interesting extraction using genetic algorithm, obtains enhancing image；
Using principal component analysis algorithm and Zero-phase Component Analysis albefactions algorithms to the enhancing image
Carry out Data Dimensionality Reduction to handle with de-redundancy, obtain dimensionality reduction and the matrix image after whitening processing.
2. according to claim 1, a kind of new finger vena preprocess method, it is characterised in that described to original graph
Comprise the following steps as carrying out finger areas extraction process：
1) gray processing processing is carried out to the original image using rgb2gray, obtains two-dimensional matrix image after gray processing, wherein
Rgb2gray formula are as follows：
Gray=0.3R+0.59G+0.11B
Wherein Gray represents grey, and R represents red, and G represents green, and B represents blueness；
2) Edge extraction processing is carried out to two-dimensional matrix image after the gray processing using sobel edge detection operators, obtained
Take 0,1 bianry image；
3) contour extraction processing is carried out to described 0,1 bianry image using two values matrix contours extract scheme, obtains contours extract
0,1 bianry image afterwards；
4) branch process is removed to 0,1 bianry image after the contours extract, obtains image after removal branch；
5) finger train of thought extension processing is carried out to the incomplete finger edge of image after the removal branch, obtains after train of thought extends and scheme
Picture, image is image after region of interesting extraction after the train of thought extends.
3. according to claim 2, a kind of new finger vena preprocess method, it is characterised in that described to use two-value
Matrix contours extract scheme carries out extraction process to described 0,1 bianry image to be included：
Described 0,1 bianry image is retrieved since the image upper left corner, runs into the value for 1, is started to detect whether as full curve,
If full curve then records coordinate in Matrix C, so as to be connected by the fraction area of absence filled in discontinuous curve
Continuous curve, i.e. finger edge profile, obtain profile matrix C, C comprising two road wheel profiles and represent as follows：
C=[C (1) C (2) ... C (k) ... C (N)]
The form of wherein each contour line represents as follows：
C (k)=[level1 x1 x2 x3 ... numxy y1 y2 y3 ...]
Wherein level1 is brightness value grade, and numxy is profile point number, each (x, y) determine on profile a bit (x,
Y), thus obtain contours extract after 0,1 bianry image.
4. according to right wants 2, a kind of new finger vena preprocess method, it is characterised in that described to described 0,1 two
Value image, which is removed branch process, to be included：
To the horizontal direction of described 0,1 bianry image as base, set up one along described 0,1 bianry image vertical direction and vertical visit
Survey line, described 0,1 bianry image from left to right record refer to the intersection point number countA of vein edge image and vertical detector with
And the number in the range of three pixels of vertical line or so, if number is more than presetting threshold value, what vertical detector passed through
Number is set as 0 entirely in the range of pixel and the pixel of left and right three；
To described 0,1 bianry image vertical direction as base, level detection line is set up along described 0,1 bianry image horizontal direction,
Described 0,1 bianry image contains pixel point value in the range of recording level detection three pixels of line and horizontal line or so from top to bottom
For 1 number, it is determined as edge if number exceedes presetting threshold value, is otherwise branch, pixel point value is all put in branch
For 0.
5. according to right wants 2, a kind of new finger vena preprocess method, it is characterised in that described to the removal
The incomplete finger edge of image after branch, which carries out the extension of finger train of thought, to be included：
To the image after the removal branch since middle part with detecting vertical curve, run into discrete point then determined level
Pixel number in line adjacent domain, if pixel number exceedes pre-set threshold, with the horizontal level of surrounding pixel point and
Coordinate of the average of vertical position as current discontinuity point；
If four corners of the image after the removal branch are free of continuity point, the angle of inclination that record refers to vein edge is θ,
It is as follows if the upper right corner, calculation formula：
Y (f (i+tan (θ/360*2* π) * (k-j)), k)=1
Wherein, y represents image array, and f represents bracket function, and (i, j) represents pixel, and k represents to wait the coordinate for extending point.
6. according to claim 1, a kind of new finger vena preprocess method, it is characterised in that described using heredity
Algorithm carries out image enhancement processing to image after the region of interesting extraction and comprised the following steps：
1) chromosome structure is initialized, produces one group of n random number, for the scope of random number between 0 to 255, n represents chromosome
Size, represent the number of gray value grade in the extracted region images, random number sequence and first element be set to
0, nth elements are set to 255；
2) corresponding picture number of individuals is randomly generated, as the first generation, the chromosome element of the first generation is obtained by the first step；
3) edge of the extracted region images is calculated using fitness function, the fitness numerical value of each chromosome is obtained, presses
According to ranking fitness individual；
4) present age is selected and bred, according to fitness by wheel disc method assortative mating individual, breeding includes mating and dashed forward
Become, produce individual of future generation and form colony, wherein what is do not mated is direct to the next generation；
5) the 4) step is repeated, until meeting that end condition reaches iterations, records the chromosome structure finally given；
6) chromosome structure is mapped using tonal gradation mapping, obtains contrast enhancing image, the contrast enhancing
Image is input picture.
A kind of 7. according to claim 1, new finger vena preprocess method, it is characterised in that it is described using it is main into
Constituent analysis algorithm and Zero-phase Component Analysis albefactions algorithms carry out Data Dimensionality Reduction to the enhancing image
Comprise the following steps with de-redundancy processing：
1) the enhancing picture size is normalized to 40*80, then cutting is 10 × 10 fritter, and sliding window moves every time
5 units, and each fritter is preserved into vector；
2) each vector is averaged, then each element subtracts average in vector, and the vector of acquisition is designated as x；
3) sigma value is obtained, calculation formula is as follows：
<mrow>
<mi>&Sigma;</mi>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>m</mi>
</mfrac>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>m</mi>
</msubsup>
<mrow>
<mo>(</mo>
<msup>
<mi>x</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>)</mo>
</mrow>
<msup>
<mrow>
<mo>(</mo>
<msup>
<mi>x</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>)</mo>
</mrow>
<mi>T</mi>
</msup>
</mrow>
4) the eigenvectors matrix U and eigenvalue matrix S of ∑ are calculated using singular value decomposition, each row of matrix U include a spy
Sign vector, S is diagonal matrix, the characteristic vector of same number of columns in each value homography U on its diagonal, vector
XPCAwhite is that the vectorial calculation formula after dimensionality reduction is as follows：
XPCAwhite=diag (1./sqrt (diag (S)+epsilon)) * U'*x
Wherein, epsilon is constant, is set as 3*10-5；
5) vector x ZCAwhite is calculated, i.e. dimensionality reduction and the vector after whitening processing, calculation formula is as follows：
XZCAwhite=U*xPCAwhite
To the further solution of formula：
XZCAwhite=U*diag (1./sqrt (diag (S)+epsilon)) * U'*x
Wherein, U is the characteristic vector obtained by step 4), and S is the matrix that diagonal includes characteristic value, and x is input data, is led to
Acquisition xZCAwhite is crossed, a sub-picture is transformed to the matrix image of a secondary size reduction.
8. a kind of new finger vena pretreatment system, it is characterised in that the system includes：
Area image extraction module：For carrying out finger areas extraction process to original image, after obtaining region of interesting extraction
Image；
Image enhancement module：For image after the region of interesting extraction to be carried out into image enhancement processing, enhancing image is obtained；
Image dimensionality reduction and whitening processing module：Handled for the enhancing image to be carried out into Data Dimensionality Reduction with de-redundancy, obtain drop
Dimension and the matrix image after whitening processing.
A kind of 9. according to claim 8, new finger vena pretreatment system, it is characterised in that the area image
Extraction mould includes：
Image gray processing processing unit：For the original image to be carried out into gray processing processing, two-dimensional matrix after gray processing is obtained
Image；
Edge extraction processing unit：For two-dimensional matrix image after the gray processing to be carried out into Edge extraction processing,
Obtain 0,1 bianry image；
Image outline extraction process unit：For described 0,1 bianry image to be carried out into contour extraction processing, after obtaining contours extract
0,1 bianry image；
Image debranching enzyme processing unit：For 0,1 bianry image after the contours extract to be removed into branch process, acquisition is gone
Except image after branch；
Finger train of thought extends processing unit：Prolong for the incomplete finger edge of image after the removal branch to be carried out into finger train of thought
Long processing, obtain image after train of thought extends.
10. according to claim 8, a kind of new finger vena pretreatment system, it is characterised in that described image strengthens
Module includes：
Chromosome initialization unit：For producing one group of n random number, for the scope of random number between 0 to 255, n represents dyeing
The size of body, the number of gray value grade in image after the region of interesting extraction is represented, to random number sequence and first
Individual element is set to 0, and nth elements are set to 255；
Generation chromosome acquiring unit：For randomly generating corresponding picture number of individuals, as the first generation, the chromosome of the first generation
Element is obtained by the first step；
Chromosome fitness computing unit：For calculating the edge of the extracted region images, the adaptation of each chromosome is obtained
Number of degrees value, according to ranking fitness individual；
Individual choice and diaspore：For being selected the present age and being bred, wheel disc method assortative mating is passed through according to fitness
Individual, breeding include mating and mutation, produce individual of future generation and form colony, wherein directly arriving for not mated is next
Generation；
Chromosome structure map unit：For mapping chromosome structure, contrast enhancing image is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710656229.4A CN107563293A (en) | 2017-08-03 | 2017-08-03 | A kind of new finger vena preprocess method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710656229.4A CN107563293A (en) | 2017-08-03 | 2017-08-03 | A kind of new finger vena preprocess method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107563293A true CN107563293A (en) | 2018-01-09 |
Family
ID=60973971
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710656229.4A Pending CN107563293A (en) | 2017-08-03 | 2017-08-03 | A kind of new finger vena preprocess method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107563293A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111079551A (en) * | 2019-11-25 | 2020-04-28 | 五邑大学 | Finger vein identification method and device based on singular value decomposition and storage medium |
| CN111079756A (en) * | 2018-10-19 | 2020-04-28 | 杭州萤石软件有限公司 | Method and equipment for extracting and reconstructing table in document image |
| CN112784837A (en) * | 2021-01-26 | 2021-05-11 | 电子科技大学中山学院 | Region-of-interest extraction method and device, electronic equipment and storage medium |
| CN112863165A (en) * | 2021-01-14 | 2021-05-28 | 深圳市子瑜杰恩科技有限公司 | Logistics enterprise fleet management method and system based on 5G |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1522953A1 (en) * | 2003-10-07 | 2005-04-13 | Sony Corporation | Image matching method, program, and system |
| CN104123706A (en) * | 2014-08-11 | 2014-10-29 | 徐州工程学院 | Image enhancement method based on adaptive immunity genetic algorithm |
-
2017
- 2017-08-03 CN CN201710656229.4A patent/CN107563293A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1522953A1 (en) * | 2003-10-07 | 2005-04-13 | Sony Corporation | Image matching method, program, and system |
| CN104123706A (en) * | 2014-08-11 | 2014-10-29 | 徐州工程学院 | Image enhancement method based on adaptive immunity genetic algorithm |
Non-Patent Citations (2)
| Title |
|---|
| 付文: "基于深度学习的MRI前列腺分割", 《中国优秀硕士学位论文全文数据库 信息科技辑（电子期刊）》 * |
| 庞晓红: "指静脉身份识别算法的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑（电子期刊）》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111079756A (en) * | 2018-10-19 | 2020-04-28 | 杭州萤石软件有限公司 | Method and equipment for extracting and reconstructing table in document image |
| CN111079756B (en) * | 2018-10-19 | 2023-09-19 | 杭州萤石软件有限公司 | Form extraction and reconstruction method and equipment in receipt image |
| CN111079551A (en) * | 2019-11-25 | 2020-04-28 | 五邑大学 | Finger vein identification method and device based on singular value decomposition and storage medium |
| CN111079551B (en) * | 2019-11-25 | 2023-09-05 | 五邑大学 | Finger vein recognition method and device based on singular value decomposition and storage medium |
| CN112863165A (en) * | 2021-01-14 | 2021-05-28 | 深圳市子瑜杰恩科技有限公司 | Logistics enterprise fleet management method and system based on 5G |
| CN112784837A (en) * | 2021-01-26 | 2021-05-11 | 电子科技大学中山学院 | Region-of-interest extraction method and device, electronic equipment and storage medium |
| CN112784837B (en) * | 2021-01-26 | 2024-01-30 | 电子科技大学中山学院 | Region of interest extraction method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105574063B (en) | The image search method of view-based access control model conspicuousness | |
| Lin | Face detection in complicated backgrounds and different illumination conditions by using YCbCr color space and neural network | |
| CN104408449B (en) | Intelligent mobile terminal scene literal processing method | |
| CN109376611A (en) | A kind of saliency detection method based on 3D convolutional neural networks | |
| WO2018145470A1 (en) | Image detection method and device | |
| CN104866616B (en) | Monitor video Target Searching Method | |
| CN113537008B (en) | Micro expression recognition method based on self-adaptive motion amplification and convolutional neural network | |
| CN107563293A (en) | A kind of new finger vena preprocess method and system | |
| CN106934386B (en) | A kind of natural scene character detecting method and system based on from heuristic strategies | |
| JP2000003452A (en) | Method for detecting face surface in digital picture, its detecting device, picture judging method, picture judging device and computer readable record medium | |
| CN101615245A (en) | Expression recognition method based on AVR and enhancing LBP | |
| Shivakumara et al. | A new multi-modal approach to bib number/text detection and recognition in Marathon images | |
| CN111667400B (en) | Human face contour feature stylization generation method based on unsupervised learning | |
| CN109063619A (en) | A kind of traffic lights detection method and system based on adaptive background suppression filter and combinations of directions histogram of gradients | |
| CN103996046A (en) | Personnel recognition method based on multi-visual-feature fusion | |
| CN107392105B (en) | Expression recognition method based on reverse collaborative salient region features | |
| Mohamed et al. | Face detection based neural networks using robust skin color segmentation | |
| CN106874848A (en) | A kind of pedestrian detection method and system | |
| CN114332559B (en) | RGB-D significance target detection method based on self-adaptive cross-mode fusion mechanism and deep attention network | |
| Gui et al. | A fast caption detection method for low quality video images | |
| CN107679467A (en) | A kind of pedestrian's weight recognizer implementation method based on HSV and SDALF | |
| Kumar et al. | Fish disease detection using HOG and FAST feature descriptor | |
| CN109800758A (en) | A kind of natural scene character detecting method of maximum region detection | |
| CN104680123A (en) | Object identification device, object identification method and program | |
| CN112395919A (en) | Off-line signature authentication method and system based on multi-feature metric learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180109 |
|
| WD01 | Invention patent application deemed withdrawn after publication |