CN105574865B - Based on the method for improving ant group algorithm extraction eyelashes - Google Patents
Based on the method for improving ant group algorithm extraction eyelashes Download PDFInfo
- Publication number
- CN105574865B CN105574865B CN201510936749.1A CN201510936749A CN105574865B CN 105574865 B CN105574865 B CN 105574865B CN 201510936749 A CN201510936749 A CN 201510936749A CN 105574865 B CN105574865 B CN 105574865B
- Authority
- CN
- China
- Prior art keywords
- eyelashes
- human oasis
- oasis exploited
- exploited
- human
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- 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)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to iris recognition technology field more particularly to a kind of methods for extracting eyelashes based on improvement ant group algorithm, and implement as follows: (1) eyelashes extract the initialization of area information element；A human oasis exploited is put, algorithm parameters are set；(2) every human oasis exploited successively executes step (3) and step (4)；(3) position of human oasis exploited selection next step, termination condition are to cover defined step number s or at the end of one's rope；(4) Pheromone update is carried out to the position that human oasis exploited passes through；(5) after all human oasis exploited traversals, the pheromone concentration of image position has been obtained, has then extracted eyelashes edge using OTSU algorithm；(6) obtained eyelashes edge is integrated, obtains complete eyelashes；(7) noise is eliminated using sequence sliding window, obtains final eyelashes.When human oasis exploited spacing takes 25 or so pixel, algorithm not only can guarantee the speed of eyelash detection but also can guarantee the effect of eyelash detection the present invention.
Description
Technical field
The invention belongs to iris recognition technology field more particularly to a kind of methods for extracting eyelashes based on improvement ant group algorithm.
Background technique
Iris recognition is due to having many advantages, such as generality, uniqueness, stability, non-infringement property, so being considered current
One of most potential biological feather recognition method.The main function of eyelashes is to enter eyes for foreign matters such as blocks dusts,
However when extracting iris feature, eyelashes there may come a time when to some extent to block iris formation, to affect true iris
The extraction of feature.Therefore, in the pretreatment stage of iris recognition, accurately detect that eyelashes occlusion area is to guarantee iris recognition
The very important link of accuracy rate.
Include: at present Kong Waikin etc. about the main method of eyelash detection, eyelashes are divided into dispersion and aggregation two
Class is done convolution using one-dimensional Gabor filter and image to dispersion eyelashes, if result is less than given threshold value, is judged as
Eyelashes；For assembling the detection of eyelashes, then the small rectangular window of a 5*5 is used, if the gray scale difference of window is less than default threshold
Value, then the center of the rectangular window can be determined as an eyelashes point；Huang Junzhou etc. is mentioned according to phase equalization first
The marginal information for taking noise positions eyelash region then in conjunction with the information in edge and region；Suhad A etc. is first to image
Greyscale transformation is carried out to enhance the contrast of image, eyelashes are then detected according to given threshold value using soble operator；Walid
Aydi etc. finds out the diagonal gradient of image first, then extracts eyelashes by preset threshold；Yuan Weiqi et al. is proposed using more
The different method of kind extracts eyelashes point by choosing corresponding threshold value, and the eyelashes point that then these are extracted again combines
Form final eyelashes.Carry out firm et al. the method using gray scale morphology, by the image binaryzation after morphology operations, Ji Kejian
Measure eyelash.
Above method is to determine whether some pixel belongs to eyelashes using the method for artificial preset threshold mostly.This side
The advantage of method is simple, efficient, and limitation is often to be difficult to obtain due to optimal threshold, to cause the result of eyelash detection
It is not satisfactory enough.
Summary of the invention
The present invention is directed to provide a kind of in place of overcome the deficiencies in the prior art to extract eyelashes based on improving ant group algorithm
Method.The effect that this method extracts eyelashes is significantly improved compared with other related algorithms；And when human oasis exploited spacing takes 25 left sides
When right pixel, algorithm not only can guarantee the speed of eyelash detection but also can guarantee the effect of eyelash detection.
In order to solve the above technical problems, the present invention is implemented as follows:
Based on the method for improving ant group algorithm extraction eyelashes, can implement as follows:
(1) eyelashes extract the initialization of area information element；A human oasis exploited is put, algorithm parameters are set；
(2) every human oasis exploited successively executes step (3) and step (4)；
(3) position of human oasis exploited selection next step, termination condition are to cover defined step number s or at the end of one's rope；
(4) Pheromone update is carried out to the position that human oasis exploited passes through；
(5) after all human oasis exploited traversals, the pheromone concentration of image position has been obtained, OTSU is then used
Algorithm extracts eyelashes edge；
(6) obtained eyelashes edge is integrated, obtains complete eyelashes；
(7) noise is eliminated using sequence sliding window, obtains final eyelashes.
As a preferred embodiment, in step (1) of the present invention, a human oasis exploited is put in every region m*m.
Further, in step (3) of the present invention, human oasis exploited selects the position of next step according to following equation:
allowedi=0,1 ... and n-1 } indicate the position for allowing selection within the scope of ant i eight neighborhood in next step；Wherein τk
Indicate the pheromone concentration at the k of eight neighborhood position；ηkIndicate the local direction factor, ηk=θik/dik, wherein θikIt indicates from position i
First position bigger than the region EA initial information element concentration mean value ave encountered along eight neighborhood direction to the zone boundary EA
rkThe initial information element concentration at place,
M, N respectively indicate the height and width in the region EA；dikIndicate position i and rkThe distance between；
Indicate human oasis exploited from position i along the first of eight neighborhood direction to all pixels bigger than ave value in the zone boundary EA
Beginning pheromone concentration mean value；
Parameter alpha, β, μ respectively indicate each Factor Weight.
Further, in step (4) of the present invention, information is carried out to the position that human oasis exploited passes through according to following equation
Element updates:
τk(n)=ρ τk(n-1)+Δτk(n)*ωk*vk(n)
Wherein τk(n) after indicating n-th human oasis exploited traversal, the pheromone concentration at the k of position；ρ indicates pheromones evaporation
Coefficient；Δτk(n) pheromones that n-th human oasis exploited discharges at the k of position are indicated；ωkIndicate that initial information element is dense at the k of position
The ratio between degree and the entire region EA initial information element concentration mean value.
Further, after n-th human oasis exploited traversal of the present invention, initial information element concentration and n-th artificial ant at the k of position
The ratio between all routing information element concentration mean values that ant passes through:
trailnIndicate all pixels location sets that n-th human oasis exploited passes through；N indicates trailnGather interior element
Number.
Further, it in step (6) of the present invention, uses
S (i, j)=sleft(i,j+1)+sright(i,j-1)
Wherein sleft(i, j) indicates left edge image, sright(i, j) indicates right hand edge image.
Eyelash detection is an important link of iris recognition pretreatment stage, and the present invention proposes a kind of based on improved ant
The method that group's algorithm extracts eyelashes.This method passes through the introducing inside and outside direction factor in eyelashes region first and enables human oasis exploited fast
Speed gathers eyelashes edge, and is updated by taking global and local two kinds of strategies to pheromones, then uses
OTSU algorithm segments the image into eyelashes edge and non-eyelashes edge two parts according to the pheromone concentration of gained image.Finally,
The eyelashes edge being partitioned into is integrated, except making an uproar, final eyelashes are obtained.The results showed that this method extracts eyelashes
Effect is significantly improved compared with other related algorithms；And when human oasis exploited spacing takes 25 or so pixel, algorithm can
Guarantee that the speed of eyelash detection can guarantee the effect of eyelash detection again.
Detailed description of the invention
Present invention will be further explained below with reference to the attached drawings and specific embodiments.Protection scope of the present invention not only office
It is limited to the statement of following content.
Fig. 1 is that eyelashes of the present invention extract area schematic；
Fig. 2 is comparison schematic diagram of the algorithms of different to the library CASIA-IrisV1 iris image eyelash detection effect；
Fig. 3 is algorithms of different to the comparison schematic diagram from acquisition iris image eyelash detection effect；
Fig. 4 is ant spacing and runing time relationship；
Fig. 5-1, Fig. 5-2 and Fig. 5-3 are ant spacing schematic diagram.
Specific embodiment
As shown, being implemented as follows based on the method for improving ant group algorithm extraction eyelashes:
(1) eyelashes extract the initialization of area information element；A human oasis exploited is put, algorithm parameters are set；
(2) every human oasis exploited successively executes step (3) and step (4)；
(3) position of human oasis exploited selection next step, termination condition are to cover defined step number s or at the end of one's rope；
(4) Pheromone update is carried out to the position that human oasis exploited passes through；
(5) after all human oasis exploited traversals, the pheromone concentration of image position has been obtained, OTSU is then used
Algorithm extracts eyelashes edge；
(6) obtained eyelashes edge is integrated, obtains complete eyelashes；
(7) noise is eliminated using sequence sliding window, obtains final eyelashes.
In step (1) of the present invention, a human oasis exploited is put in every region m*m.
In step (3) of the present invention, human oasis exploited selects the position of next step according to following equation:
allowedi=0,1 ... and n-1 } indicate the position for allowing selection within the scope of ant i eight neighborhood in next step；Wherein τk
Indicate the pheromone concentration at the k of eight neighborhood position；ηkIndicate the local direction factor, ηk=θik/dik, wherein θikIt indicates from position i
First position bigger than the region EA initial information element concentration mean value ave encountered along eight neighborhood direction to the zone boundary EA
rkThe initial information element concentration at place,
M, N respectively indicate the height and width in the region EA；dikIndicate position i and rkThe distance between；
Indicate human oasis exploited from position i along the first of eight neighborhood direction to all pixels bigger than ave value in the zone boundary EA
Beginning pheromone concentration mean value；
Parameter alpha, β, μ respectively indicate each Factor Weight.
In step (4) of the present invention, Pheromone update is carried out to the position that human oasis exploited passes through according to following equation:
τk(n)=ρ τk(n-1)+Δτk(n)*ωk*vk(n)
Wherein τk(n) after indicating n-th human oasis exploited traversal, the pheromone concentration at the k of position；ρ indicates pheromones evaporation
Coefficient；Δτk(n) pheromones that n-th human oasis exploited discharges at the k of position are indicated；ωkIndicate that initial information element is dense at the k of position
The ratio between degree and the entire region EA initial information element concentration mean value.
After n-th human oasis exploited traversal of the present invention, initial information element concentration is passed through with n-th human oasis exploited at the k of position
The ratio between all routing information element concentration mean values:
trailnIndicate all pixels location sets that n-th human oasis exploited passes through；N indicates trailnGather interior element
Number.
In step (6) of the present invention, use
S (i, j)=sleft(i,j+1)+sright(i,j-1)
Wherein sleft(i, j) indicates left edge image, sright(i, j) indicates right hand edge image.
The present invention proposes that a kind of new eyelashes extracting method based on improvement ant group algorithm, this method are not by artificial pre-
If the method for threshold value determines eyelashes, but human oasis exploited search eyelashes edge is first passed through, constantly enhanced during search
The information at eyelashes edge, while weakening non-edge information, optimal eyelashes segmentation threshold is then automatically found by OTSU algorithm again.
This ensures that preferable eyelashes segmentation effect.
Ant group algorithm, which is Italy scholar Dorigo M in 1991 et al., to be obtained during the ant colony foraging behavior
It inspires, a kind of simulated evolutionary algorithm of simulation ant behavior of proposition, we are known as Ant Algorithm (AS), due to the algorithm
Have the characteristics that information positive feedback, distribution calculates and heuristic search, has been successfully applied to path optimization in recent years,
Fault diagnosis and image segmentation.
The human oasis exploited that can be understood as simulation ant behavior to the process that iris image carries out eyelashes extraction is looked for food process.
Image is regarded as two-dimensional matrix, an element in each location of pixels homography, put in every n*n pixel region one it is artificial
Ant.Since the characteristics of eyelashes is that grey scale pixel value is lower and edge gradient is higher, according to this feature, the search of human oasis exploited
Target is exactly the pixel that gray value is small and gradient is big.In order to improve the precision of eyelashes extraction, part is introduced inside eyelashes region
Direction factor, the factor guide human oasis exploited close to the higher eyelashes of closer and pheromone concentration from him；It adopts respectively simultaneously
Global policies and local policy is taken to be updated pheromones, so that the pheromone concentration at eyelashes edge is further increased
By force.In this way by several step iteration after, the pheromone concentration at eyelashes edge will significantly be higher than the letter of non-eyelashes fringe region
Plain concentration is ceased, eyelashes edge can be easily partitioned by being then split further according to OTSU algorithm to obtained image.
Finally, being integrated to the eyelashes edge being partitioned into, except making an uproar, final eyelashes have just been obtained.Steps are as follows by the present invention:
The determination in 3.1 eyelashes extraction region
Find according to observing a large amount of iris images: the range that the eyelashes of common people block iris is the above area, pupil center
Domain, so, it is in image using the center of pupil as baseline that we, which select eyelashes to extract the region (eyelash abstract, EA),
Above section, as shown in figure 1 shown in the above dash area of heavy black.
The determination of 3.2 eyelashes extraction area information element initial value
In order to improve the search efficiency of algorithm, the pheromones initial value τ of imagek(0) it is arranged as follows.
graykIndicate the grey scale pixel value at the k of position, gradkIndicate the pixel gradient value at the k of position, expression formula such as formula
(2) shown in:
gradk=| f (i+1, j)-f (i, j) |+| f (i, j+1)-f (i, j) | (2)
Wherein f (i, j) indicates the grey scale pixel value at k.In order to exclude noise influence, it is specified that image gradient be less than set
Fixed threshold gamma
In order to consistent with gradient, the gray scale of image is also normalized to (0~γ).
In this way, then pheromones initial value τk(0) range is (0~1).
In addition, will form to human oasis exploited Path selection dry since the gradient at pupil edge is larger while gray scale is lower
It disturbs, so setting 0 to the pheromones initial value of pupil and its edge environ.
The definition of 3.3 transition probabilities
Human oasis exploited to be reached within the scope of eight neighborhood according to transition probability selection in next step every time from current location i
Position k.Herein, transition probability formula is defined as follows:
allowedi={ 0,1 ... n-1 } indicates next step in ant i eight neighborhood to allow the position of selection that (ant is
Position through accessing does not allow reselection).
Wherein τkIndicate the pheromone concentration at the k of eight neighborhood position.
ηkIndicate the local direction factor, ηk=θik/dik, wherein θikIt indicates from position i along eight neighborhood direction to the region EA
First position r bigger than the region EA initial information element concentration mean value ave that boundary is encounteredkThe initial information element concentration at place,
M, N respectively indicate the height and width in the region EA.dikIndicate position i and rkThe distance between.When human oasis exploited is walked
When inside into eyelashes region, parameter ηijSo that human oasis exploited when searching for eyelashes edge, that is, considers the higher side of pheromone concentration
Edge direction, while the lower edge direction being but closer of pheromone concentration is had also contemplated, to ensure that the essence that eyelashes extract
Degree.
Indicate human oasis exploited from position i along the first of eight neighborhood direction to all pixels bigger than ave value in the zone boundary EA
Beginning pheromone concentration mean value.Because eyelashes marginal information element concentration is higher than ave, the bigger direction of mean value is exactly to exist
The bigger direction of eyelashes possibility.When human oasis exploited is when eyelashes region exterior, which enables human oasis exploited quick
The direction existing for the eyelashes is close.
Parameter alpha, β, μ respectively indicate each Factor Weight, currently, the setting of ant group algorithm relevant parameter is still without theoretically
Foundation, the setting of these parameters determines mainly by experience.
3.4 pheromone update strategy
After every human oasis exploited covers defined step number, the pheromones of position are adjusted according to formula (7):
τk(n)=ρ τk(n-1)+Δτk(n)*ωk*vk(n) (7)
τ in formula (7)k(n) after indicating n-th human oasis exploited traversal, the pheromone concentration at the k of position.ρ indicates that pheromones are steamed
Send out coefficient, Δ τk(n) pheromones that n-th human oasis exploited discharges at the k of position are indicated, are constants.ωkIndicate initial at the k of position
The ratio between pheromone concentration and the entire region EA initial information element concentration mean value, ωkIt is worth bigger explanation at the k of position relative to the overall situation
Pheromone concentration is bigger.
In order to further increase the pheromone concentration of target, pheromones local updating strategy is also used, is introduced parameter v (n)k
vk(n) after indicating n-th human oasis exploited traversal, initial information element concentration is passed through with n-th human oasis exploited at the k of position
The ratio between all routing information element concentration mean values.trailnIndicate all pixels location sets that n-th human oasis exploited passes through.N
Indicate trailnGather the number of interior element.vk(n) in all pixels position that n-th human oasis exploited of the bigger explanation of value passes through, position
The generic pixel element concentration set at k is bigger.
After n-th human oasis exploited traverses, the pheromones increment at the k of position is according to ωkAnd vk(n) size is proportional
Be adjusted correspondingly, the higher Messages element of such pheromones increases faster, the lower Messages element of pheromones
What is increased is slower, to be conducive to further increase eyelashes extraction effect.
The segmentation of 3.5 eyelashes
During all ants are according to direction transition strategy and pheromone update strategy traversal image, eyelashes edge
Pheromone concentration can be increasing, and after all ants have traversed, the pheromone concentration at eyelashes edge generally will obviously compare
Non- eyelashes marginal information element concentration is high.At this moment, then using OTSU algorithm image is split, so that it may easily be partitioned into
Eyelashes edge.
The integration of 3.6 eyelashes
Since the target of ant group algorithm search is the pixel that gradient is big and gray scale is small, and there are certain width for eyelashes itself
Degree, so what is extracted using method as above is that intermediate there are the eyelashes in gap.Complete eyelashes in order to obtain, need to gap
It is filled.Binaryzation is carried out to the image after segmentation, edge pixel gray value sets 255, and non-edge pixels gray value is 0.Root
The intermediate gap width of discovery, which is observed, according to many experiments approximately is less than 3 pixels.In order to remove intermediate gap, extract complete
Eyelashes, we find out the image at the left and right edge of target two-value eyelashes image respectively, since the characteristics of eyelashes gray scale is intermediate
Low both sides are high, so, the method for discrimination of left edge is: the position is edge (bianry image gray value 255) and original gradation
The position right pixels gray value is centainly lower than current location in image；It is side that similarly the method for discrimination of right hand edge, which is the position,
The position leftmost pixel gray value is centainly than current location in edge (bianry image gray value 255) and original-gray image
It is low.Then final eyelashes are shown using following formula.
S (i, j)=sleft(i,j+1)+sright(i,j-1) (10)
Wherein sleft(i, j) indicates left edge image, sright(i, j) indicates right hand edge image.Because there are overlapping phenomenon,
So being restored to 255 according to formula (11) to the gray value of the pixel of overlapping.
3.7 eliminate noise using sequence sliding window
The eyelashes for taking algorithm as above to extract can have some noises, in order to further improve eyelashes extraction effect, need
Remove and makes an uproar.Discovery eyelashes are usually continuous according to the observation, and these noises are some isolated spots.What we took
Method is to eliminate noise using sequence sliding window.The side length of series of windows is 3~L pixel, and thickness window is 1 pixel, first
First, image is slipped over using the window of 3*3 pixel-by-pixel, in sliding process, if 8 location of pixels of surrounding are without gray value
255 pixel, and the pixel that intermediate pixel is 255, it is determined that the pixel is noise, its value is changed to 0.Then, successively make
The noise surrounded by window is eliminated using same method with the window that side length is 4~L.Attention window selection cannot be too big, such as
Fruit is excessive not only to increase program runtime, and eyelashes is mistakened as doing noise elimination sometimes, to affect eyelashes extraction
Precision；If selection is too small, some noises cannot be removed.Experience have shown that: L takes 5 ideal.
CASIA-IrisV1 iris picture library and this research department oneself of Institute of Automation, CAS is respectively adopted in present invention experiment
Acquisition iris picture library is verified.Wherein iris image resolution ratio in the library CASIA-IrisV1 is 320*280, this research department acquires certainly
Iris image resolution ratio is 800*600.The dominant frequency for testing service machine is 2.71GHz, inside saves as 2G, operating system is
Windows xp, programming used tool are Visual C++2010.Due to (WANIKIN K, the DAVID Z.Detecting of document 1
eyelash and reflection for accurate iris segmentation[J].International
Journal of Pattern Recognition and Artificial Intelligence,2003,17(6):1025-
And (AYDI W, KAMOUN L, the MASMOUDI N.A Fast and Accurate Eyelids and of document 4 1034)
Eyelashes Detection Approach for Iris Segmentation[J].Journal of Multimedia
Processing and Technologies, 2012,3 (4): 166-173.) both for no rainbow for carrying out standardization processing
Film image.So the two algorithms is selected to compare with the present invention.In addition, document 1 and document 4 can all generate when extracting eyelashes
Noise, in order to objectively be compared algorithm, both algorithms are also the method used as this paper to the filtering of noise.
Table 1 provides the typical set-up of major parameter of the present invention.
The setting of 1 algorithm parameter of table
The extraction effect of 4.1 eyelashes compares
As can be seen from Figure 2: Fig. 2 (a) is original image；Fig. 2 (b) is document [1] algorithm；Fig. 2 (c) is document [4] algorithm；
Fig. 2 (d) is the present invention.There are many phenomenon that interrupting in the eyelashes that Fig. 2 (b), Fig. 2 (c) algorithm extract, this is mainly due to same
The gray value of root eyelashes different parts is often different, and caused by decision threshold is constant.In contrast, figure of the present invention
2 (d) the eyelashes continuitys extracted are significantly improved than algorithm pattern 2 (b), Fig. 2 (c).Meanwhile it may also be seen that:
The region (such as eyelashes upper left) closeer for eyelashes, other opposite algorithms, the effect that Fig. 2 (d) of the present invention is extracted also relatively are managed
Think.Still there is the end of some eyelashes to extract in Fig. 2 (d) imperfect, be primarily due to the gray value of the end of these eyelashes
It is closer to background, and gradient value is smaller, so the more difficult identification of human oasis exploited.
As shown in Figure 3: Fig. 3 (a) is original image；Fig. 3 (b) is document [1] algorithm；Fig. 3 (c) is document [4] algorithm；Fig. 3
It (d) is the present invention.The present invention is to this research department from the detection effect of iris image eyelashes is acquired also superior to other algorithms.In Fig. 3
(d) there is also several places, and accidentally the patch edge on iris image to be come out as eyelash detection in, and this is mainly due to these patches
Edge feature very close to eyelashes the characteristics of, so causing erroneous judgement.In addition, due to the visible light iris picture color compared with
Secretly, the gray value of gray value and eyelashes is very close, this also results in part eyelashes and certain parts of eyelashes and can not detect
Come.
The relationship of 4.2 ant spacing and runing time and eyelash detection effect
In order to analyze the relationship of runing time Yu human oasis exploited spacing, an artificial ant is put at different pixels spacing respectively
Ant obtains the average operating time of algorithm and the relationship of human oasis exploited spacing as shown in figure 4, as shown in Figure 4: human oasis exploited spacing
It is bigger, Riming time of algorithm is shorter；And human oasis exploited spacing is smaller, Riming time of algorithm is longer.When human oasis exploited spacing is less than
When 20 or so pixel, the runing time of program is very fast as the increase of human oasis exploited spacing declines；And when between human oasis exploited
When away from 20 or so pixel is greater than, program runtime decline is slower；
In addition, it can be seen from the experiment that: human oasis exploited spacing is bigger, and the effect that eyelashes extract is poorer；Spacing is smaller, extraction effect
Better, by taking Fig. 2 (a) as an example, Fig. 5-1, Fig. 5-2 and Fig. 5-3 are to use different interval number human oasis exploited using the present invention respectively
Extract runing time and operational effect that eyelashes are spent.It can be seen that Fig. 5-1 be ant spacing: 5 pixels, runing time:
11.273s；Fig. 5-2 is ant spacing, 15 pixels, runing time: 1.502s；Fig. 5-3 is ant spacing: 40 pixels, when operation
Between: 0.458s.Fig. 5-1 and Fig. 5-2 effect difference is not obvious, and apparent fracture then occur in the eyelashes that Fig. 5-3 is extracted.
Factor of both tradeoff as above simultaneously passes through experimental verification repeatedly: taking 25 left sides when choosing human oasis exploited skip number n
When right pixel, preferable eyelashes extraction effect not only can guarantee, but also keep the runing time of calling program shorter.
It is to be understood that being merely to illustrate the present invention above with respect to specific descriptions of the invention and being not limited to this
Technical solution described in inventive embodiments, those skilled in the art should understand that, still the present invention can be carried out
Modification or equivalent replacement, to reach identical technical effect；As long as meet use needs, all protection scope of the present invention it
It is interior.
Claims (1)
1. based on the method for improving ant group algorithm extraction eyelashes, which is characterized in that implement as follows:
(1) eyelashes extract the initialization of area information element；A human oasis exploited is put in every m*m pixel region；
(2) every human oasis exploited successively executes step (3) and step (4)；
(3) position of human oasis exploited selection next step, termination condition are step number s as defined in human oasis exploited is covered or can without road
It walks；
(4) Pheromone update is carried out to the position that human oasis exploited passes through according to following equation:
τk(n)=ρ τk(n-1)+Δτk(n)*ωk*vk(n)；
Wherein τk(n) after indicating n-th human oasis exploited traversal, the pheromone concentration at the k of position；ρ indicates pheromones evaporation coefficient；
Δτk(n) pheromones that n-th human oasis exploited discharges at the k of position are indicated；ωkIndicate position k at initial information element concentration with it is whole
A eyelashes extract the ratio between region initial information element concentration mean value；vk(n) initial at the k of position after indicating n-th human oasis exploited traversal
The ratio between all routing information element concentration mean values that pheromone concentration and n-th human oasis exploited pass through；
(5) after all human oasis exploited traversals, the pheromone concentration of the image position of human oasis exploited process has been obtained, so
Eyelashes edge is extracted using OTSU algorithm afterwards；
(6) obtained eyelashes edge is integrated, obtains complete eyelashes；
(7) noise is eliminated using sequence sliding window, obtains final eyelashes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510936749.1A CN105574865B (en) | 2015-12-14 | 2015-12-14 | Based on the method for improving ant group algorithm extraction eyelashes |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510936749.1A CN105574865B (en) | 2015-12-14 | 2015-12-14 | Based on the method for improving ant group algorithm extraction eyelashes |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105574865A CN105574865A (en) | 2016-05-11 |
| CN105574865B true CN105574865B (en) | 2019-11-12 |
Family
ID=55884956
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510936749.1A Expired - Fee Related CN105574865B (en) | 2015-12-14 | 2015-12-14 | Based on the method for improving ant group algorithm extraction eyelashes |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105574865B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108171201B (en) * | 2018-01-17 | 2021-11-09 | 山东大学 | Rapid eyelash detection method based on gray scale morphology |
| CN109919963B (en) * | 2019-03-14 | 2023-03-24 | 吉林大学 | Vehicle paint defect position detection method |
| CN112241722A (en) * | 2020-11-18 | 2021-01-19 | 河南工业大学 | Antarctic sea ice remote sensing image segmentation method based on ant colony algorithm |
| CN112446871B (en) * | 2020-12-02 | 2022-11-15 | 山东大学 | Tunnel crack identification method based on deep learning and OpenCV |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1885314A (en) * | 2006-07-11 | 2006-12-27 | 电子科技大学 | Pre-processing method for iris image |
| US20080025575A1 (en) * | 2004-03-22 | 2008-01-31 | Microsoft Corporation | Iris-Based Biometric Identification |
| CN101201893A (en) * | 2006-09-30 | 2008-06-18 | 电子科技大学中山学院 | Iris recognizing preprocessing method based on grey level information |
| CN101241550A (en) * | 2008-01-19 | 2008-08-13 | 电子科技大学中山学院 | Iris image quality judging method |
| CN101339603A (en) * | 2008-08-07 | 2009-01-07 | 电子科技大学中山学院 | Method for selecting qualified iris image from video frequency stream |
| US20100014718A1 (en) * | 2008-04-17 | 2010-01-21 | Biometricore, Inc | Computationally Efficient Feature Extraction and Matching Iris Recognition |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8098901B2 (en) * | 2005-01-26 | 2012-01-17 | Honeywell International Inc. | Standoff iris recognition system |
-
2015
- 2015-12-14 CN CN201510936749.1A patent/CN105574865B/en not_active Expired - Fee Related
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080025575A1 (en) * | 2004-03-22 | 2008-01-31 | Microsoft Corporation | Iris-Based Biometric Identification |
| CN1885314A (en) * | 2006-07-11 | 2006-12-27 | 电子科技大学 | Pre-processing method for iris image |
| CN101201893A (en) * | 2006-09-30 | 2008-06-18 | 电子科技大学中山学院 | Iris recognizing preprocessing method based on grey level information |
| CN101241550A (en) * | 2008-01-19 | 2008-08-13 | 电子科技大学中山学院 | Iris image quality judging method |
| US20100014718A1 (en) * | 2008-04-17 | 2010-01-21 | Biometricore, Inc | Computationally Efficient Feature Extraction and Matching Iris Recognition |
| CN101339603A (en) * | 2008-08-07 | 2009-01-07 | 电子科技大学中山学院 | Method for selecting qualified iris image from video frequency stream |
Non-Patent Citations (16)
| Title |
|---|
| A Novel Eyelash Detection Method for Iris Recognition;Weiqi Yuan et al;《Proceedings of the 2005 IEEE Engineering in Medicine and Biology 27th Annual Conference》;20050901;6536-6539 * |
| Accurate Iris Segmentation Based on Novel Reflection and Eyelash Detection Model;W.K.Kong et al;《Proceedings of 2001 International Symposium on Intelligent Multimedia, Mdeo and Speech Processing》;20010502;263-266 * |
| Decision Level Fusion of Iris and Signature Biometrics for Personal Identification using Ant Colony Optimization;Ibrahim A. Saleh et al;《International Journal of Engineering and Innovative Technology (IJEIT)》;20140531;第3卷(第11期);35-42 * |
| Eyelid and Eyelash Segmentation Based on Wavelet Transform for Iris Recognition;Shahram Javadi et al;《2011 4th International Congress on Image and Signal Processing》;20111231;1231-1235 * |
| Indian Iris Recognition System using Ant Colony Optimization;Anupam Tiwari et al;《International Journal of Engineering Trends and Technology (IJETT)》;20150430;第21卷(第8期);380-387 * |
| 一种有效抑制睫毛干扰的虹膜定位算法;唐荣年 等;《西安交通大学学报》;20071031;第41卷(第10期);1175-1178 * |
| 一种有效的虹膜定位及睫毛检测方法;陈广华 等;《计算机工程与应用》;20100630;第46卷(第6期);171-173,207 * |
| 一种用于虹膜识别的眼睑和睫毛检测算法;周俊 等;《武汉理工大学学报·信息与管理工程版》;20110430;第33卷(第2期);175-179 * |
| 一种用于虹膜识别的眼睫毛遮挡检测算法;苑玮琦 等;《光电工程》;20080630;第35卷(第6期);124-129 * |
| 一种高效的睫毛及眼睑检测方法;常乐 等;《微电子学与计算机》;20110430;第28卷(第4期);122-126,130 * |
| 基于虹膜识别的眼睫毛检测;苑玮琦 等;《微计算机信息》;20081231;第24卷(第10-1期);311-312,295 * |
| 基于蚁群和Hough变换的虹膜定位算法;康景磊 等;《计算机科学》;20121130;第39卷(第11A期);384-385,394 * |
| 基于蚁群算法的改进Otsu理论的图像多阈值分割;王爽 等;《微计算机应用》;20080430;第29卷(第4期);25-28 * |
| 形态学算子和小波变换的虹膜去噪算法;郭业才 等;《数据采集与处理》;20130930;第28卷(第5期);586-590 * |
| 用于虹膜识别的眼睑及眼睫毛遮挡检测;来毅 等;《计算机辅助设计与图形学学报》;20070331;第19卷(第3期);346-350 * |
| 虹膜分割中眼睑和睫毛的检测;罗忠亮 等;《石河子大学学报（自然科学版）》;20090630;第27卷(第3期);379-382 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105574865A (en) | 2016-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105574865B (en) | Based on the method for improving ant group algorithm extraction eyelashes | |
| Li et al. | Integrating holistic and local deep features for glaucoma classification | |
| Kande et al. | Unsupervised fuzzy based vessel segmentation in pathological digital fundus images | |
| Sun et al. | Fast linear feature detection using multiple directional non‐maximum suppression | |
| CN108416344A (en) | Eyeground color picture optic disk and macula lutea positioning identifying method | |
| CN105761260B (en) | A kind of skin image affected part dividing method | |
| Mittal et al. | Automated detection and segmentation of drusen in retinal fundus images | |
| Abdelsamea | An automatic seeded region growing for 2d biomedical image segmentation | |
| Alghmdi et al. | Measurement of optical cup-to-disc ratio in fundus images for glaucoma screening | |
| Vimal et al. | RETRACTED ARTICLE: A method of progression detection for glaucoma using K-means and the GLCM algorithm toward smart medical prediction | |
| CN101866420A (en) | Image preprocessing method for optical volume holographic iris recognition | |
| Kavitha et al. | An approach to identify optic disc in human retinal images using ant colony optimization method | |
| Zhao et al. | Automatic identification and morphometry of optic nerve fibers in electron microscopy images | |
| Sharma et al. | Dynamic thresholding technique for detection of hemorrhages in retinal images | |
| Ganjee et al. | A novel microaneurysms detection method based on local applying of Markov random field | |
| CN108230306A (en) | Eyeground color picture blood vessel and arteriovenous recognition methods | |
| Sae-Tang et al. | Exudates detection in fundus image using non-uniform illumination background subtraction | |
| Li et al. | An automated method using hessian matrix and random walks for retinal blood vessel segmentation | |
| Murugan et al. | An automatic localization of microaneurysms in retinal fundus images | |
| Fahimuddin et al. | Retinal boundary segmentation in OCT images using active contour model | |
| Helwan | ITDS: Iris tumor detection system using image processing techniques | |
| Huang et al. | A local adaptive algorithm for microaneurysms detection in digital fundus images | |
| Tamilarasi et al. | Template matching algorithm for exudates detection from retinal fundus images | |
| González et al. | Cost function selection for a graph-based segmentation in OCT retinal images | |
| Das et al. | Entropy thresholding based microaneurysm detection in fundus images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20191112 Termination date: 20211214 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |