CN108470171B - The asynchronous coding comparison method of two dimension - Google Patents
The asynchronous coding comparison method of two dimension Download PDFInfo
- Publication number
- CN108470171B CN108470171B CN201810838874.2A CN201810838874A CN108470171B CN 108470171 B CN108470171 B CN 108470171B CN 201810838874 A CN201810838874 A CN 201810838874A CN 108470171 B CN108470171 B CN 108470171B
- Authority
- CN
- China
- Prior art keywords
- iris
- region
- confidence region
- registration
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
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
- G06V40/18—Eye characteristics, e.g. of the iris
- G06V40/19—Sensors therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/18—Eye characteristics, e.g. of the iris
- G06V40/193—Preprocessing; Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Ophthalmology & Optometry (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention discloses a kind of asynchronous coding comparison methods of two dimension, have not only considered the rotational differential of pixel, but also take into account the fault-tolerance of iris inside and outside circle segmentation error, realize the asynchronous comparison of two dimension of the iris pixel in angular direction and radial direction.Compared with traditional one-dimensional pixel comparison method（Only consider iris angular direction shift calibrating）, the present invention considers the segmentation error during iris pixel.Compared with synchronous two-dimensional pixel compares（Correction is moved integrally in radial direction and angular direction）, the alignments of this method are more flexible.This method not only covers the method for synchronizing two dimension comparison, and two-dimensional pixel is allowed to occur the comparison for misplacing and being overlapped in reasonable range, therefore the present invention can preferably adapt to the center of circle offset and radius error occurred during iris segmentation, it is small with required calculation amount, the fast advantage of detection speed.
Description
Technical field
The present invention relates to iris recognition technology field, more particularly to a kind of asynchronous coding comparison method of two dimension.
Background technology
The textural characteristics of iris random distribution are that the mankind are natural, identity of high confidence level, thus iris recognition at
It is science and technology and the trend of market development for important one of authentication means.
Iris recognition process generallys include iris segmentation, and iris space coordinate is converted to polar coordinates, iris texture enhancing, rainbow
Film pixel and pixel ratio pair.The invention mainly relates to content be iris recognition during pixel compare link.
Iris is the organ for adjusting pupil size, and front is viewed as ring-type, and the inside and outside circle of iris respectively refers to pupil
The outer edge of outer edge and iris；The difference of rotation angle when traditional pixel ratio only considered pixel to (Daugman methods)
There is no consideration, when dividing iris inside and outside circle, there is also certain errors, or have ignored this fractional error.However in fact,
Inside and outside circle segmentation error be inevitable, especially using circle fitting method in, do not consider speckle noise etc. interference because
Element, shape get over the iris inside and outside circle of stray circle model, and the uncertainty of segmentation is bigger.Such as since perspective deformation or camera lens are abnormal
Become, the iris image that iris inner circle is partial to ellipse, the pupil for being partial to ellipse for one, segmentation may be collected
As a result it may be partial to elliptical arbitrary one side, it is also possible to occupy elliptical center.Due to pixel it is stringent depend on inside and outside circle
Segmentation result, thus plausible segmentation error may result in erroneous judgement in some cases.It still uses above-mentioned
The example of iridovalosis, the result of segmentation, which may be partial to the most fenestra of fitting points, may be partial to an average circle,
It is either way reasonable from the point of view of fitting, but during space coordinate is converted to polar coordinates, expansion
Image might have prodigious difference, and this difference also results in the difference of pixel.In this case if only considering rotation
On difference and ignore segmentation error be clearly unreasonable.Iridovalosis herein is an example, practical application
In even a stringent circular pupil also inevitably exist segmentation error, this error include center location error and
The error of radius of circle.
Invention content
In order to solve the above-mentioned technical problem, not only consider the rotational differential of pixel, but also take into account iris inside and outside circle segmentation error
Fault-tolerance, the present invention propose the asynchronous coding comparison method of two dimension.
The asynchronous coding comparison method of two dimension includes iris registration and iris recognition, and the iris registration includes step
Suddenly：
10, the first iris image that acquisition is registered for iris first, and divide first iris image and obtain first
Annular iris area image；First annular iris area image transformation is expanded into the first rectangular image；
11, normalization obtains registered images template, and the registered images template is divided into several regions；
12, several regions are divided into the first confidence region C1, the second confidence region C2, third confidence region
C3；
The iris recognition includes step：
20, second iris image of the acquisition for iris recognition, and divide second iris image and obtain the second ring-type
Iris region image；Second annular iris area image transformation is expanded into the second rectangular image；
21, normalization obtains images to be recognized；
22, all first confidence region C1 in registered images template are taken, are registrated with images to be recognized；
23, it is set with second according to the first confidence region C1 registration results and enrollment the first confidence region C1
The position relationship of reliability region C2 takes in registered images template all second confidence region C2 in images to be recognized corresponding position
Place is registrated with images to be recognized；
24, according to the registration result of the first confidence region C1 and the second confidence region C2, figure to be identified is calculated
As the matching score with registered images template.
The asynchronous coding comparison method of two dimension of the invention further improvement lies in that, each area size is identical, two
The cut-off rule of a adjacent area is parallel with the vertical edge of the first rectangular image.
The asynchronous coding comparison method of two dimension of the invention further improvement lies in that, the step S22 takes registered images mould
All first confidence region C1 in plate, the step of being registrated with images to be recognized include：First by all first confidence levels
Region C1 is traversed in all positions for meeting registration position condition, is obtained the first confidence region of registered images template C1 and is existed
The registration score of each position, to be registrated the highest position of score as registration result.
The asynchronous coding comparison method of two dimension of the invention further improvement lies in that, the registration position condition includes：It will
The first or second rectangular image transverse direction edge is known as angular direction, is denoted as the directions x, and vertical edge direction is known as radial direction, is denoted as y
Direction；Offset between the first adjacent confidence region C1 on the direction x, y is less than or equal to the second maximum offset；It is all
Offset between first confidence region C1 and images to be recognized in y-direction is less than or equal to the first maximum offset；Not phase
Between the first adjacent confidence region C1 in the distance in the directions x in the first preset range.
The asynchronous coding comparison method of two dimension of the invention further improvement lies in that, first maximum offset is 5
Pixel.
The asynchronous coding comparison method of two dimension of the invention further improvement lies in that, first preset range is registration mould
Relative displacement between non-conterminous first confidence region C1 described in plate in the x direction is less than 5 degree of corresponding pixel numbers.
The asynchronous coding comparison method of two dimension of the invention further improvement lies in that, registered images mould is taken described in step S23
All second confidence region C2 include the step of images to be recognized corresponding position is registrated with images to be recognized in plate：
The second confidence region C2 that adjacent region has been registrated is taken, is traversed in all positions for meeting registration position condition, with
The highest position of score is registrated as registration result, until all second confidence region C2 obtain registration result completion and match
It is accurate.
The asynchronous coding comparison method of two dimension of the invention further improvement lies in that, the registration position condition includes：It will
The first or second rectangular image transverse direction edge is known as angular direction, is denoted as the directions x, and vertical edge direction is known as radial direction, is denoted as y
Direction；Offset between adjacent region on the direction x, y is less than or equal to the second maximum offset, each in registered images template
Iris region is less than the first maximum offset relative to the offset of images to be recognized in y-direction.
The asynchronous coding comparison method of two dimension of the invention further improvement lies in that, second maximum offset is 1
Pixel.
It is non-in the two dimension of angular direction and radial direction that the asynchronous coding comparison method of two dimension of the invention realizes iris pixel
It is synchronous to compare.Compared with traditional one-dimensional pixel comparison method（Only consider iris angular direction shift calibrating）, present invention consideration
Segmentation error during iris pixel.Compared with synchronous two-dimensional pixel compares（It is i.e. whole in radial direction and angular direction
Shift calibrating）, the alignments of this method are more flexible.This method not only covers the method for synchronizing two dimension comparison, Er Qieyun
Perhaps there is the comparison for misplacing and being overlapped in reasonable range in two-dimensional pixel, therefore the present invention can preferably adapt to iris segmentation mistake
The center of circle offset occurred in journey and radius error.
Consider that the time loss problem that two-dimensional pixel compares, the present invention are divided and are based on by using iris region confidence level
The scoring algorithm of confidence level improves computational efficiency.Assuming that KY indicates that radial direction allows mobile maximum distance, KX to indicate angle
Direction allows mobile maximum distance, N to be iris region number.If compared using common two dimension, zoning scoring
Number be（2*KX+1）*（2*KY+1）* N.And for the present invention, it is assumed that high confidence region iris region number is Nc1, in
Confidence region iris region number is Nc2, and the number of Nc1+Nc2≤N, zoning scoring are（2*KX+1）*（2*
KY+1）* Nc1 + 9 * Nc2.Its calculation scale is less than common two-dimensional pixel and compares.To sum up, the present invention has taken into account pixel ratio
To computational accuracy and calculate the time, have preferable technique effect.
Description of the drawings
Iris registration in Fig. 1 the method for the invention and iris recognition flow chart.
Fig. 2 is iris region division and expanded schematic diagram.
Specific implementation mode
Below in conjunction with the accompanying drawings and specific implementation mode present invention is further described in detail.
The asynchronous coding comparison method of two dimension of the invention, asynchronous is for synchronization.Synchronizing will be entire
Iris region is seen as an entirety, and synchronous comparison method is that integral translation compares in angular direction and radial direction.It is asynchronous
Method on condition that be each region by iris region segmentation, occur misplacing and be overlapped between allowing each region to see.
As shown in Figure 1, the asynchronous coding comparison method of present invention two dimension includes iris registration and iris recognition, the iris
Registration includes step：
10, the first iris image that acquisition is registered for iris first, and divide first iris image and obtain first
Annular iris area image；First annular iris area image transformation is expanded into the first rectangular image；Fig. 2 shows this
Kind transformational relation, Fig. 2 includes annular iris area image schematic diagram 21, rectangular image schematic diagram 23 and iris image zone number
First annular iris area image is placed in rectangular coordinate system by schematic diagram 22 specifically, initially setting up rectangular coordinate system；So
It converts the rectangular co-ordinate of pixel in the first annular iris area image to polar coordinates afterwards, is finally unfolded first under polar coordinates
Annular iris area image obtains the first rectangular image.
11, normalization obtains registered images template, and the registered images template is divided into several wide regions；
Preferably, the registered images template is divided into 16 regions in the present embodiment, as shown in Figure 2；Preferably, in the present embodiment
The angle of the corresponding iris inner circle in each region is π/8 degree；It is described normalization refer to refer to that series of standards has been carried out to image
Processing transformation, be allowed to be transformed to the process of a fixed standard form.
12, several regions are divided into the first confidence region C1, the second confidence region C2, third confidence region
C3；The confidence region was divided according to the case where under normal circumstances, each region is blocked by eyelid, it is preferred that by
S15 shown in 21, s0, s7 in Fig. 2, it is minimum that s8 (s indicate section, that is, iris region) by eyelid is blocked probability, is
The region that iris maximum possible goes out cruelly, therefore s15, s0, s7, s8 are divided into first confidence level area in the present embodiment
Domain C1；Similarly, due to s3, s4, s10, s11, s12, the maximum probability that s13 is blocked by eyelid, be iris exposure most
Low region, therefore in the present embodiment by s3, s4, s10, s11, s12, s13 are divided into first confidence region
C3；Remaining s14, s1, s2, s5, s6, s9 are as the second confidence level/centre confidence region C2.
The iris recognition includes step：
20, second iris image of the acquisition for iris recognition, and divide second iris image and obtain the second ring-type
Iris region image；Second annular iris area image transformation is expanded into the second rectangular image；Transformation rule and step 10
It is identical.
21, normalization obtains images to be recognized.
22, all first confidence region C1 in registered images template are taken, are registrated with images to be recognized：Step packet
It includes：All first confidence region C1 are traversed in all positions for meeting registration position condition first, obtain registration figure
Registration score as template the first confidence region C1 in each position is tied using being registrated the highest position of score as registration
Fruit.The registration position condition includes：The first or second rectangular image transverse direction edge is known as angular direction, is denoted as the directions x,
Vertical edge direction is known as radial direction, is denoted as the directions y；Offset between the first adjacent confidence region C1 on the direction x, y
Less than or equal to the second maximum offset, the offset refer to two regions on the basis of the relative position of acquiescence in the x-direction or y
The pixel number that direction is moved into line misregistration or overlapping；Between first confidence region C1 and images to be recognized in y-direction
Offset is less than or equal to the first maximum offset；In the distance in the directions x first between non-conterminous first confidence region C1
In preset range.First maximum offset is 5 pixels；First preset range is non-conterminous described in enrollment
The first confidence region C1 between relative displacement in the x direction be less than 5 degree of corresponding pixel numbers, it is assumed that the square after normalization
Shape image horizontal edge totally 128 pixels, since rectangular image horizontal edge corresponds to round totally 360 degree of angular direction, 5 degree of corresponding pixels
Number is 1.77,2 pixels of rounding；Second maximum offset is 1 pixel；The pixel refers to the image after normalization
Pixel；Rectangular image schematic diagram is as shown in 23 in Fig. 2；The registration refers to that template area is similar to images to be recognized region
Degree and matching degree highest；Assuming that two dimensional encoded images to be compared are A and B, matching degree is as follows to calculation：
Formula 1
Formula 2
Wherein k indicates that the index of iris region, sk indicate that k-th of iris region, i expressions are the i-th row of pixel image, j
Indicate that the jth row of pixel image, α k are the distances that k-th of iris region of integer representation is moved in radial direction, KY indicates radius
Direction allows mobile maximum distance, β k to be the distances that k-th of iris region of integer representation is moved in angular direction, and KX indicates angle side
Indicate that exclusive or, #0 indicate 0 number to mobile maximum distance, XOR is allowed；In order to ensure the continuity of image, adjacent son
Module is required in angular direction and radial direction at most there are one the error of pixel, i.e. formula 1 also needs to meet the following conditions：
Formula 3
For minimum confidence region C3, it is believed that its influenced by eyelashes and eyelid it is more serious, thus in the present embodiment
It is not included in comparison.
23, it is set with second according to the first confidence region C1 registration results and enrollment the first confidence region C1
The position relationship of reliability region C2 takes in registered images template all second confidence region C2 in images to be recognized corresponding position
Place is registrated with images to be recognized, and step includes：The second confidence region C2 that adjacent region has been registrated is taken, is matched in satisfaction
All positions of quasi- locality condition are traversed, to be registrated the highest position of score as registration result, until all second set
Reliability region C2 obtains registration result and completes registration；The registration position condition includes：By the first or second histogram
As transverse direction edge is known as angular direction, the directions x are denoted as, vertical edge direction is known as radial direction, is denoted as the directions y；Between adjacent region
Offset on the direction x, y is less than or equal to the second maximum offset, and second maximum offset is 1 pixel；Registration
Matching degree computational methods are：
Formula 4
Since wherein the second maximum offset is 1 pixel,Value range be：
Formula 5
24, according to the registration result of the first confidence region C1 and the second confidence region C2, figure to be identified is calculated
As the matching score with registered images template, i.e.,
Formula 6
It is using normalized score：
Formula 7
Wherein H indicates height/line number of two-dimensional pixel image, and Ws is the width of an iris region.
It is described the invention in detail above in association with accompanying drawings and embodiments, those skilled in the art can basis
Above description makes many variations example to the present invention.Thus, certain details in embodiment should not constitute limitation of the invention,
The present invention will be using the range that the appended claims define as protection scope of the present invention.
Claims (6)
1. the asynchronous coding comparison method of two dimension, including iris registration and iris recognition, it is characterised in that the iris registration packet
Include step：
10, the first iris image that acquisition is registered for iris first, and divide first iris image and obtain the first ring-type
Iris region image；First annular iris area image transformation is expanded into the first rectangular image；
11, normalization obtains registered images template, and the registered images template is divided into several regions；
12, several regions are divided into the first confidence region C1, the second confidence region C2, third confidence region C3；
The iris recognition includes step：
20, second iris image of the acquisition for iris recognition, and divide second iris image and obtain the second annular iris
Area image；Second annular iris area image transformation is expanded into the second rectangular image；
21, normalization obtains images to be recognized；
22, all first confidence region C1 in registered images template are taken, are registrated with images to be recognized, step includes：It is first
All first confidence region C1 are traversed in all positions for meeting registration position condition first, obtain registered images template
First confidence region C1 each position registration score, to be registrated the highest position of score as registration result；
23, it is set with second according to the first confidence region C1 registration results and registered images template the first confidence region C1
The position relationship of reliability region C2 takes in registered images template all second confidence region C2 in images to be recognized corresponding position
Place is registrated with images to be recognized, and step includes：The second confidence region C2 that adjacent region has been registrated is taken, is matched in satisfaction
All positions of quasi- locality condition are traversed, to be registrated the highest position of score as registration result, until all second set
Reliability region C2 obtains registration result and completes registration；
24, according to the registration result of the first confidence region C1 and the second confidence region C2, be calculated images to be recognized with
The matching score of registered images template.
2. the asynchronous coding comparison method of two dimension according to claim 1, it is characterised in that every in several regions
A equal size in region is identical, and the cut-off rule of two adjacent areas is parallel with the vertical edge of the first rectangular image.
3. the asynchronous coding comparison method of two dimension according to claim 2, it is characterised in that described in step 22 and step 23
Registration position condition includes：The first or second rectangular image transverse direction edge is known as angular direction, is denoted as the directions x, vertical edge side
To referred to as radial direction, it is denoted as the directions y；The adjacent region of any two in the confidence region that registered images template is registrated
Between offset on the direction x, y be less than or equal to the second maximum offset；What is be registrated in registered images template all sets
Offset between reliability region and images to be recognized in y-direction is less than or equal to the first maximum offset；Registered images template
Between non-conterminous first confidence region C1 in the distance in the directions x in the first preset range.
4. the asynchronous coding comparison method of two dimension according to claim 3, it is characterised in that first maximum offset
It is 5 pixels.
5. the asynchronous coding comparison method of two dimension according to claim 3, it is characterised in that first preset range is
Relative displacement between non-conterminous first confidence region C1 described in registered images template in the x direction is less than 5 degree of correspondences
Pixel number.
6. the asynchronous coding comparison method of two dimension according to claim 3, it is characterised in that second maximum offset
It is 1 pixel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810838874.2A CN108470171B (en) | 2018-07-27 | 2018-07-27 | The asynchronous coding comparison method of two dimension |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810838874.2A CN108470171B (en) | 2018-07-27 | 2018-07-27 | The asynchronous coding comparison method of two dimension |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108470171A CN108470171A (en) | 2018-08-31 |
| CN108470171B true CN108470171B (en) | 2018-11-02 |
Family
ID=63259894
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810838874.2A Active CN108470171B (en) | 2018-07-27 | 2018-07-27 | The asynchronous coding comparison method of two dimension |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108470171B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109785349A (en) * | 2019-01-07 | 2019-05-21 | 哈尔滨理工大学 | A kind of polar coordinates edge coding method based on circle model |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100374708B1 (en) * | 2001-03-06 | 2003-03-04 | 에버미디어 주식회사 | Non-contact type human iris recognition method by correction of rotated iris image |
| CN1271559C (en) * | 2004-06-15 | 2006-08-23 | 沈阳工业大学 | Human iris identifying method |
| CN100373396C (en) * | 2006-06-27 | 2008-03-05 | 电子科技大学 | Iris identification method based on image segmentation and two-dimensional wavelet transformation |
| CN101154265A (en) * | 2006-09-29 | 2008-04-02 | 中国科学院自动化研究所 | Method for recognizing iris with matched characteristic and graph based on partial bianry mode |
-
2018
- 2018-07-27 CN CN201810838874.2A patent/CN108470171B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN108470171A (en) | 2018-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109544447B (en) | Image splicing method and device and storage medium | |
| US9774837B2 (en) | System for performing distortion correction and calibration using pattern projection, and method using the same | |
| CN103426186B (en) | A kind of SURF fast matching method of improvement | |
| EP2678824B1 (en) | Determining model parameters based on transforming a model of an object | |
| US8155395B2 (en) | Iris authentication apparatus | |
| US20110075933A1 (en) | Method for determining frontal face pose | |
| David et al. | Object recognition in high clutter images using line features | |
| CN104091155B (en) | The iris method for rapidly positioning of illumination robust | |
| CN111340701B (en) | Circuit board image splicing method for screening matching points based on clustering method | |
| CN102844768B (en) | The shielding of image template | |
| US9117145B2 (en) | Finger biometric sensor providing coarse matching of ridge flow data using histograms and related methods | |
| CN103400384A (en) | Large viewing angle image matching method capable of combining region matching and point matching | |
| CN102722887A (en) | Image registration method and device | |
| US20210150181A1 (en) | Image acquisition system for off-axis eye images | |
| CN108470171B (en) | The asynchronous coding comparison method of two dimension | |
| Jang et al. | A study on eyelid localization considering image focus for iris recognition | |
| CN103679672A (en) | Panorama image splicing method based on edge vertical distance matching | |
| CN107292272A (en) | A kind of method and system of the recognition of face in the video of real-time Transmission | |
| CN102004911A (en) | Method for improving accuracy of face identification | |
| CN110097587A (en) | Robust structure light pattern for 3D camera system | |
| Liu et al. | Improved rectangle template matching based feature point matching algorithm | |
| CN117911668A (en) | Drug information identification method and device | |
| US8724888B2 (en) | Stereo vision based dice recognition system and stereo vision based dice recognition method for uncontrolled environments | |
| KR100786204B1 (en) | Deformation-resilient iris recognition methods | |
| CN103136754B (en) | A kind of image blurring direction discrimination method of feature based Block direction differential |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |