CN106203375A - A kind of based on face in facial image with the pupil positioning method of human eye detection - Google Patents
A kind of based on face in facial image with the pupil positioning method of human eye detection Download PDFInfo
- Publication number
- CN106203375A CN106203375A CN201610570904.7A CN201610570904A CN106203375A CN 106203375 A CN106203375 A CN 106203375A CN 201610570904 A CN201610570904 A CN 201610570904A CN 106203375 A CN106203375 A CN 106203375A
- Authority
- CN
- China
- Prior art keywords
- image
- face
- human eye
- pupil
- region
- 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
- 238000000034 method Methods 0.000 title claims abstract description 134
- 210000001747 pupil Anatomy 0.000 title claims abstract description 51
- 238000001514 detection method Methods 0.000 title claims abstract description 40
- 230000001815 facial effect Effects 0.000 title claims abstract description 25
- 230000000877 morphologic effect Effects 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims abstract description 7
- 210000001508 eye Anatomy 0.000 claims description 90
- 230000011218 segmentation Effects 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 2
- 238000005260 corrosion Methods 0.000 claims description 2
- 230000007797 corrosion Effects 0.000 claims description 2
- 235000008733 Citrus aurantifolia Nutrition 0.000 claims 1
- 235000011941 Tilia x europaea Nutrition 0.000 claims 1
- 239000004571 lime Substances 0.000 claims 1
- 238000012360 testing method Methods 0.000 abstract description 4
- 238000006243 chemical reaction Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 230000002708 enhancing effect Effects 0.000 description 3
- 210000003128 head Anatomy 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 210000005252 bulbus oculi Anatomy 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 210000000720 eyelash Anatomy 0.000 description 1
- 210000000744 eyelid Anatomy 0.000 description 1
- VIKNJXKGJWUCNN-XGXHKTLJSA-N norethisterone Chemical compound O=C1CC[C@@H]2[C@H]3CC[C@](C)([C@](CC4)(O)C#C)[C@@H]4[C@@H]3CCC2=C1 VIKNJXKGJWUCNN-XGXHKTLJSA-N 0.000 description 1
- 238000013386 optimize process Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/18—Eye characteristics, e.g. of the iris
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/02—Affine transformations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30041—Eye; Retina; Ophthalmic
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Ophthalmology & Optometry (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of based on face in facial image with the pupil positioning method of human eye detection, including training face grader and human eye grader, utilize the detection of face classification device and extract human face region image, utilize human eye detection of classifier after human face region image is carried out pretreatment and extract human eye area image, human eye area image is carried out binaryzation and Morphological scale-space, the method using region projection reduces detection range, uses centroid method to carry out accurate pupil center point location.Application is the present invention test facial image, and result shows, can reduce hunting zone during Pupil diameter, improves accuracy and the speed of Pupil diameter.
Description
Technical field
The present invention relates to a kind of pupil positioning method, be specifically related to a kind of based on face in facial image and human eye detection
Pupil positioning method.
Background technology
The authentication of living things feature recognition has proved to be a kind of effective method and verifies the identification and more of a people
The password that safety ratio is traditional, password, or the hardware mankind based on token identify system.At present, there is several biological identification technology, as
The geometry iris identification of face, fingerprint, iris and hands has been demonstrated that in independent research be bio-identification the most accurately, and attracts
Substantial amounts of concern.
Additionally, tracking technique, being referred to as eye tracking technology, be to use electronics, machinery, optics and other detection meanss obtain
Obtain the current state of the eyes of user, then analyze the current eye position of human eye technology.The most most popular based on video
Method.This method is by installing anteorbital video camera, the pupil image of real-time capture eyes.Then, the party
Method can calculate the current eye position of human eye, it is achieved thereby that use sight line and the mutual purpose of computer.
Therefore, pupil center location is an important research direction in iris identification and eye tracking field, is eyeball
Motion rotates the first step of research.Due to the geometric properties that pupil is unique, the circle of a standard can be regarded as, the most more hold
Easily training pupil target.It addition, research method also can be more rich, more accurately with stable.At present, there are many algorithms to determine pupil
Center, wherein centroid method is one of simplest method.First, pupil image is converted into bianry image by it, thus may be used
To split other parts of pupil and image easily.Using pupil barycenter as the center of pupil.Additionally, utilize edge fitting
Principle carry out the extraction of pupil boundary, then carry out justifying or the matching of ellipse.Finally, the center of the circle detected is as pupil
The center in hole.Centroid method calculates simple, and amount of calculation is little, good stability, can be realized by hardware.It is applicable to simple Background
Picture, for more complicated image, the significant reduction of accuracy of the method.Edge fitting based on pupil image rule, if pupil
Hole pattern picture is incomplete, and the accuracy of experiment will be greatly affected.
Summary of the invention
It is fixed that the technical problem to be solved in the present invention is to provide a kind of pupil based on face in facial image and human eye detection
Method for position, this pupil positioning method novelty apply face and human eye detection, region projection carry out the size of downscaled images, make
Obtain the information the most only comprising pupil in region interested, finally use centroid method to demarcate in pupil for this region
The heart, thus improve accuracy and the speed of Pupil diameter.
For solving above-mentioned technical problem, the technical solution used in the present invention is:
A kind of based on face in facial image with the pupil positioning method of human eye detection, comprise the steps:
1.1. training face grader and human eye grader；
1.2. face original image is gathered；
1.3. utilize the face classification device in step 1.1 that the face original image gathered in step 1.2 is detected, carry
Take out human face region image；
1.4. the human face region image extracted through step 1.3 is carried out Image semantic classification；
1.5. utilize the human eye grader in step 1.1 that the image after step 1.4 processes is detected, extract people
Eye area image；
1.6. the human eye area image obtained through step 1.5 is carried out binaryzation and Morphological scale-space, uses region projection
Method reduce detection range, carry out the coarse positioning of human eye；
1.7. centroid method is used to carry out accurate pupil center point location the image after step 1.6 processes.
In the technique scheme of the present invention, application face and human eye grader carry out image detection, extract corresponding
Face and human eye area image, the human eye area image extracted is carried out binaryzation and Morphological scale-space, and uses region
The method of projection reduces detection range so that the most only comprise the information of pupil in region interested, finally for this
One region uses centroid method to demarcate pupil center, thus improves accuracy and the speed of Pupil diameter.
As the further improvement of technique scheme, described face classification device and human eye grader for utilizing
The cascade classifier with haar feature that AdaBoost algorithm builds.The construction method of described cascade classifier is:
3.1. the haar feature of facial image sample set is extracted；
3.2. AdaBoost Algorithm for Training is used to go out some Weak Classifiers based on haar feature, by described some weak typings
Device is configured to strong classifier, and this strong classifier has higher classification capacity；
3.3. by some by step 3.2 obtain strong classifier cascade obtain cascade classifier.
The haar feature of face and human eye is introduced in AdaBoost algorithm, progressively can choose from substantial amounts of feature
Excellent haar feature, utilizes the face with haar feature that AdaBoost algorithm builds and human eye cascade classifier to be possible not only to
Reduce search time, improve search efficiency, but also be avoided that the interference of ambient noise.
As the further improvement of technique scheme, the Image semantic classification in step 1.4 comprises the steps:
4.1. according to the facial characteristics of face, the method using " three five, front yards ", by original for the face obtained in step 1.2
Image part at this rectangular base 1/2 at away from the top 1/8 of Face detection rectangle, as new search model
Enclose；
4.2. the new hunting zone obtained step 4.1 carries out image enhaucament, removes noise spot present in image；
4.3. the human eye area image obtained through step 4.2 process is carried out gray processing process, obtain gray level image；
4.4. the image processed through step 4.3 is smoothed；
4.5. the image processed through step 4.4 is carried out histogram equalization process.
Reduced hunting zone by above-mentioned steps further, image is optimized process.
As the further improvement of technique scheme, the image binaryzation in step 1.6 processes and includes:
5.1. the method utilizing iteration determines threshold value T of binaryzation；
5.2. the pixel value of each pixel in the image of extraction is compared with threshold value T, then will be worth accordingly and turn
Changing 0 or 255 into, when a pixel value is equal to or more than threshold value T, the numerical value at this is converted to 255；Otherwise, be converted to 0, as
Shown in formula (1.1):
In formula (1.1), p (x, y) be point (x, y) pixel value that place is to be converted, g (x, y) be in gray level image point (x,
Y) pixel value at place, T represents threshold value.
So-called iterative method is based on the thought approached, and its step is as follows:
6.1. obtain maximum gradation value and the minimum gradation value of image, be designated as Z respectivelyMAXAnd ZMIN, make initial threshold T0=
(ZMAX+ZMIN)/2；
6.2. according to threshold value TK(K >=0) divides the image into as foreground and background, obtains both average gray value Z respectivelyO
And ZB；
6.3. new threshold value T is obtainedK+1=(ZO+ZB)/2；
If 6.4. TK=TK+1, then gained is threshold value T；Otherwise go to (2nd) step, continue to calculate.
The eye image extracted from original image has double attributes, and considers the special of human eye area image
Property, so have employed the mode of iteration to determine the threshold value of binaryzation.This method can distinguish target and background well, and
Compared with former method, substantially increase processing speed.
As the further improvement of technique scheme, the Morphological scale-space in step 1.6 includes: the image to binaryzation
Use the opening operation method in morphology, eliminate less target, and discontinuous region is preferably separated, then smooth
The object boundary of large area, shown in described opening operation method such as formula (1.2):
Wherein, containing expanding and corrosion, as shown in formula (1.3) and formula (1.4) in the formula of opening operation:
Effectively filter out noise by above-mentioned process, retain the original information of image.
As the further improvement of technique scheme, step 1.6 use the method for region projection reduce detection range,
The coarse positioning carrying out human eye includes:
9.1. be averaged in the horizontal and vertical directions segmentation by eyes window, it is assumed that vertical segmentation and horizontal segmentation
Number is respectively m, n, a width of w in the region of horizontal direction after segmentation, a height of h of vertical direction, then the upright projection in the i-th region with
The floor projection function in jth region is respectively as shown in formula (1.5), (1.6):
In formula, pv(x) and phY () is illustrated respectively in the upright projection of xth column direction and in the floor projection of y line direction；
9.2. upright projection gray scale maximum region is i.e. eye image with the common factor in floor projection gray scale maximum region
The region that middle gray value is maximum, the region maximum from gray scale radiates out, and i.e. merges with the field of surrounding.Pass through above-mentioned steps
Pupil image can be included in as much as possible in this region, then this region is carried out next step process.
As the further improvement of technique scheme, centroid method is used to carry out accurate pupil center point location concrete
For:
Utilizing centroid method to process the image comprising pupil, the result obtained is in this facial image in pupil
Heart location point, the formula of centroid method is as shown in (1.7).
To sum up, the application of a kind of pupil positioning method novelty based on face in facial image and human eye detection of the present invention
Face and human eye detection, region projection carry out the size of downscaled images so that the most only comprise pupil in region interested
The information in hole, finally for this region use centroid method demarcate pupil center, thus improve Pupil diameter accuracy and
Speed.
Accompanying drawing explanation
The present invention is described further with detailed description of the invention below in conjunction with the accompanying drawings.
Fig. 1 is process of the present invention schematic diagram.
Fig. 2 is the original image of the embodiment of the present invention 1.
Fig. 3 is the image obtained after Fig. 2 carries out Face datection and the process of " three five, front yards " method.
Fig. 4 be the region confined in Fig. 3 is carried out gray processing process after the image that obtains.
Fig. 5 be Fig. 4 is smoothed after the image that obtains.
Fig. 6 be Fig. 5 is carried out histogram equalization after the image that obtains.
Fig. 7 be Fig. 6 is carried out human eye detection after the detection zone area image that obtains.
Fig. 8 be the human eye area detected is carried out binary conversion treatment after, the projection of the left eye region that obtains.
Fig. 9 be the human eye area detected is carried out binary conversion treatment after, the projection of the right eye region that obtains.
Figure 10 is that the method using the embodiment of the present invention 1 carries out the image after testing the Pupil diameter obtained.
Figure 11 is that the method using the embodiment of the present invention 2 carries out the image after testing the Pupil diameter obtained.
Figure 12 is that the method using the embodiment of the present invention 3 carries out the image after testing the Pupil diameter obtained.
Detailed description of the invention
Fig. 1 shows the implementation process of the specific embodiment of the invention, including training face grader and human eye grader；
Gather face original image；Use face classification device detection face original image, extract human face region image；To the face extracted
Area image carries out Image semantic classification；Use human eye grader to detect, extract human eye area image；To human eye area figure
As carrying out binaryzation and Morphological scale-space, use the method for region projection to reduce detection range, carry out human eye coarse positioning；Use weight
Heart method carries out Pupil diameter.
Embodiment 1
Fig. 2 is tested by the method using centroid method, region to strengthen method and the present invention, and this picture is positive face figure
Picture.Fig. 3 is the image obtained after Fig. 2 carries out Face datection and the process of " three five, front yards " method, and Fig. 4 is to the district confined in Fig. 3
The image that territory obtains after carrying out gray processing process, Fig. 5 be Fig. 4 is smoothed after the image that obtains, Fig. 6 is for enter Fig. 5
The image obtained after column hisgram equalization, Fig. 7 be Fig. 6 is carried out human eye detection after the detection zone area image that obtains, Fig. 8 is right
After the human eye area detected carries out binary conversion treatment, the left eye region projection obtained, Fig. 9 is to the human eye area detected
After carrying out binary conversion treatment, the right eye region projection obtained, Figure 10 is the pupil using the method for the present invention to obtain through embodiment 1
Image behind location, hole.
The coordinate utilizing the right and left eyes that the method for the present invention draws is (154,227), (266,224), average operation time
It is about 1.646 milliseconds；The right and left eyes coordinate that centroid method draws is (152,227), (270,225), and average operation time is about
1.706 millisecond；The result that region enhancing method draws is (154,227), (268,220), and average operation time is about 1.681 millis
Second.Embodiment 1 experimental result is as shown in table 1:
Table 1
| Method | Left eye coordinates | Right eye coordinate | Average time (ms) |
| The inventive method | (154,227) | (266,224) | 1.646 |
| Centroid method | (152,227) | (270,225) | 1.706 |
| Region strengthens method | (154,227) | (268,220) | 1.681 |
As shown in Table 1, at the average time-consuming aspect of computing, this method strengthens fast about 0.06 milli of method than centroid method and region
Second and 0.035 millisecond.
Embodiment 2
It is by three kinds of methods in embodiment 1, Figure 11 to be tested equally.In example 2, the left side that this method draws
The coordinate of right eye is (185,213), (266,210), and average operation time is about 1.422 milliseconds；The right and left eyes that centroid method draws
Coordinate is (184,211), (270,209), and average operation time is about 1.429 milliseconds；The result that region enhancing method draws is
(184,211), (271,210), average operation time is about 1.435 milliseconds.Embodiment 2 experimental result is as shown in table 2:
Table 2
| Method | Left eye coordinates | Right eye coordinate | Average time (ms) |
| The inventive method | (185,213) | (266,210) | 1.422 |
| Centroid method | (184.211) | (270,209) | 1.429 |
| Region strengthens method | (184,211) | (271,210) | 1.435 |
As shown in Table 2, the arithmetic speed of the method that embodiment 2 also demonstrates the present invention is faster than other two kinds of methods.
Embodiment 3
The detection picture of Figure 12 the first two embodiment therewith being used in embodiment 3 detecting is different, in embodiment 3
Face head portrait in Figure 12 has a certain degree of inclination, and this factor also have impact on the operation efficiency of the inventive method.Embodiment 3
Experimental result as shown in table 3.As shown in Table 3, the coordinate of the right and left eyes that the inventive method draws be (517,432), (687,
456), average operation time is about 3.249 milliseconds；The right and left eyes coordinate that centroid method draws is (515,432), (688,454), flat
All operation times are about 3.185 milliseconds；The result that region enhancing method draws is (517,429), (684,451), during average calculating operation
Between be about 3.234 milliseconds.The inventive method is than slow about 0.064 millisecond of the average operation time of centroid method, slower than region strengthens method
About 0.015 millisecond.Just can find out from the coordinate of the right and left eyes of three kinds of method location, right eye location left eye to be compared in Figure 12
High, say, that face head portrait has a certain degree of inclination to the left.By experiment, the inclination of these degree can affect this
The search efficiency of bright method, carries out the rule of human eye detection, typically carries out in positive face image according to AdaBoost algorithm
During human eye detection, search order is from top to bottom, from left to right, first obtains left eye coordinates so being usually, then obtains right eye seat
Mark.But during face's head portrait run-off the straight, according to search order, being first to obtain right eye coordinate, then obtain left eye coordinates, do so is just
Can greatly strengthen the time of search, affect the operation efficiency of algorithm.
Table 3
| Method | Left eye coordinates | Right eye coordinate | Average time (ms) |
| The inventive method | (517,432) | (687,456) | 3.249 |
| Centroid method | (515,432) | (688,454) | 3.185 |
| Region strengthens method | (517,429) | (684,451) | 3.234 |
The inventive method in the ordinary course of things, accelerates the speed of human eye location, improves operation efficiency.Further, since
The inventive method is to carry out the centralized positioning of pupil in facial image, so when method designs, can ignore eyelash and
The eyelid impact on eyes.
Above in conjunction with the drawings and specific embodiments and embodiment the present invention carried out further instruction, but this
Bright it is not limited to the above-described embodiment and examples, in the ken that those of ordinary skill in the art are possessed, it is also possible to
Make a variety of changes on the premise of without departing from present inventive concept.
Claims (10)
1. one kind based on face in facial image and the pupil positioning method of human eye detection, it is characterised in that comprise the steps:
1.1. training face grader and human eye grader；
1.2. face original image is gathered；
1.3. utilize the face classification device in step 1.1 that the face original image gathered in step 1.2 is detected, extract
Human face region image；
1.4. the human face region image extracted through step 1.3 is carried out Image semantic classification；
1.5. utilize the human eye grader in step 1.1 that the image after step 1.4 processes is detected, extract human eye district
Area image；
1.6. the human eye area image obtained through step 1.5 is carried out binaryzation and Morphological scale-space, uses the side of region projection
Method reduces detection range, carries out the coarse positioning of human eye；
1.7. centroid method is used to carry out accurate pupil center point location the image after step 1.6 processes.
The most according to claim 1 based on face in facial image with the pupil positioning method of human eye detection, its feature exists
In, described face classification device and human eye grader are the cascade sort with haar feature utilizing AdaBoost algorithm to build
Device.
The most according to claim 2 based on face in facial image with the pupil positioning method of human eye detection, its feature exists
In, the construction method of described cascade classifier is:
3.1. the haar feature of facial image sample set is extracted；
3.2. AdaBoost Algorithm for Training is used to go out some Weak Classifiers based on haar feature, by described some Weak Classifier structures
Cause strong classifier；
3.3. by some by step 3.2 obtain strong classifier cascade obtain cascade classifier.
The most according to claim 1 based on face in facial image with the pupil positioning method of human eye detection, its feature exists
In, the Image semantic classification in step 1.4 comprises the steps:
4.1. according to the facial characteristics of face, the method using " three five, front yards ", the face original image that will obtain in step 1.2
Part at this rectangular base 1/2 at away from the top 1/8 of Face detection rectangle, as new hunting zone；
4.2. the new hunting zone obtained step 4.1 carries out image enhaucament, removes noise spot present in image；
4.3. the human eye area image obtained through step 4.2 process is carried out gray processing process, obtain gray level image；
4.4. the image processed through step 4.3 is smoothed；
4.5. the image processed through step 4.4 is carried out histogram equalization process.
The most according to claim 1 based on face in facial image with the pupil positioning method of human eye detection, its feature exists
In, the image binaryzation in step 1.6 processes and includes:
5.1. the method utilizing iteration determines threshold value T of binaryzation；
5.2. the pixel value of each pixel in the image of extraction is compared with threshold value T, then value accordingly is converted into 0
Or 255, when a pixel value is equal to or more than threshold value T, the numerical value at this is converted to 255；Otherwise, be converted to 0, such as formula
(1.1) shown in:
P (x, y)=255, g (x, y) >=T (1.1)
P (x, y)=0, otherwise
In formula (1.1), (x is y) that (x, y) pixel value that place is to be converted, (x is y) at point (x, y) place in gray level image to g to point to p
Pixel value, T represents threshold value.
The most according to claim 5 based on face in facial image with the pupil positioning method of human eye detection, its feature exists
In, so-called iterative method is based on the thought approached, and its step is as follows:
6.1. obtain maximum gradation value and the minimum gradation value of image, be designated as Z respectivelyMAXAnd ZMIN, make initial threshold T0=(ZMAX+
ZMIN)/2；
6.2. according to threshold value TK(K >=0) divides the image into as foreground and background, obtains both average gray value Z respectivelyOAnd ZB；
6.3. new threshold value T is obtainedK+1=(ZO+ZB)/2；
If 6.4. TK=TK+1, then gained is threshold value T；Otherwise go to (2nd) step, continue to calculate.
The most according to claim 1 based on face in facial image with the pupil positioning method of human eye detection, its feature exists
In, the Morphological scale-space in step 1.6 includes: the image of binaryzation is used the opening operation method in morphology, eliminates less
Target, and discontinuous region is preferably separated, then smooths the object boundary of large area, described opening operation method
As shown in formula (1.2):
。
The most according to claim 7 based on face in facial image with the pupil positioning method of human eye detection, its feature exists
In, containing expanding and corrosion, as shown in formula (1.3) and formula (1.4) in the formula of described opening operation:
。
The most according to claim 1 based on face in facial image with the pupil positioning method of human eye detection, its feature exists
In, step 1.6 use the method for region projection reduce detection range, the coarse positioning carrying out human eye includes:
9.1. be averaged in the horizontal and vertical directions segmentation by eyes window, it is assumed that vertical segmentation and the number of horizontal segmentation
It is respectively m, n, a width of w in the region of horizontal direction after segmentation, a height of h of vertical direction, the then upright projection in the i-th region and jth
The floor projection function in region is respectively as shown in formula (1.5), (1.6):
In formula, pv(x) and phY () is illustrated respectively in the upright projection of xth column direction and in the floor projection of y line direction；
9.2. upright projection gray scale maximum region is i.e. that eye image is mellow lime with the common factor in floor projection gray scale maximum region
The region that angle value is maximum, the region maximum from gray scale radiates out, and i.e. merges with the field of surrounding, by pupil image as far as possible
Be included in this region.
The most according to claim 1 based on face in facial image with the pupil positioning method of human eye detection, its feature exists
In, use centroid method carry out accurate pupil center point location particularly as follows:
Utilizing centroid method to process the image comprising pupil, the result obtained is the centre bit of pupil in this facial image
Putting a little, the formula of centroid method is as shown in (1.7).
。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610570904.7A CN106203375A (en) | 2016-07-20 | 2016-07-20 | A kind of based on face in facial image with the pupil positioning method of human eye detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610570904.7A CN106203375A (en) | 2016-07-20 | 2016-07-20 | A kind of based on face in facial image with the pupil positioning method of human eye detection |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106203375A true CN106203375A (en) | 2016-12-07 |
Family
ID=57494491
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610570904.7A Pending CN106203375A (en) | 2016-07-20 | 2016-07-20 | A kind of based on face in facial image with the pupil positioning method of human eye detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106203375A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107301391A (en) * | 2017-06-16 | 2017-10-27 | 广州市百果园信息技术有限公司 | Area determination method and device, storage medium |
| WO2018108124A1 (en) * | 2016-12-15 | 2018-06-21 | 腾讯科技（深圳）有限公司 | Method and system for positioning pupil |
| CN108498064A (en) * | 2018-03-19 | 2018-09-07 | 中国人民解放军空军航空医学研究所 | Digitize bright spot scintillation detection system |
| CN108921010A (en) * | 2018-05-15 | 2018-11-30 | 北京环境特性研究所 | A kind of pupil detection method and detection device |
| CN109190509A (en) * | 2018-08-13 | 2019-01-11 | 阿里巴巴集团控股有限公司 | A kind of personal identification method, device and computer readable storage medium |
| CN110415252A (en) * | 2018-04-26 | 2019-11-05 | 北京连心医疗科技有限公司 | A kind of eye circumference organ segmentation method, equipment and storage medium based on CNN |
| CN110598635A (en) * | 2019-09-12 | 2019-12-20 | 北京大学第一医院 | Method and system for face detection and pupil positioning in continuous video frames |
| CN111488845A (en) * | 2020-04-16 | 2020-08-04 | 深圳市瑞立视多媒体科技有限公司 | Eye sight detection method, device, equipment and storage medium |
| CN112070806A (en) * | 2020-09-14 | 2020-12-11 | 北京华严互娱科技有限公司 | Real-time pupil tracking method and system based on video image |
| CN112733570A (en) * | 2019-10-14 | 2021-04-30 | 北京眼神智能科技有限公司 | Glasses detection method and device, electronic equipment and storage medium |
| FR3132967A1 (en) | 2022-02-23 | 2023-08-25 | Patrick CHARREYRON | Method for generating a reconstituted image and device for generating such a reconstituted image. |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102867172A (en) * | 2012-08-27 | 2013-01-09 | Tcl集团股份有限公司 | Human eye positioning method, system and electronic equipment |
| CN102902967A (en) * | 2012-10-16 | 2013-01-30 | 第三眼（天津）生物识别科技有限公司 | Method for positioning iris and pupil based on eye structure classification |
| CN103440476A (en) * | 2013-08-26 | 2013-12-11 | 大连理工大学 | Locating method for pupil in face video |
| CN105205480A (en) * | 2015-10-31 | 2015-12-30 | 潍坊学院 | Complex scene human eye locating method and system |
-
2016
- 2016-07-20 CN CN201610570904.7A patent/CN106203375A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102867172A (en) * | 2012-08-27 | 2013-01-09 | Tcl集团股份有限公司 | Human eye positioning method, system and electronic equipment |
| CN102902967A (en) * | 2012-10-16 | 2013-01-30 | 第三眼（天津）生物识别科技有限公司 | Method for positioning iris and pupil based on eye structure classification |
| CN103440476A (en) * | 2013-08-26 | 2013-12-11 | 大连理工大学 | Locating method for pupil in face video |
| CN105205480A (en) * | 2015-10-31 | 2015-12-30 | 潍坊学院 | Complex scene human eye locating method and system |
Non-Patent Citations (7)
| Title |
|---|
| 于起峰等: "《基于图像的精密测量与运动测量》", 31 July 2002 * |
| 徐光柱等: "《实用性目标检测与跟踪算法原理及应用》", 30 April 2015 * |
| 李兰友等: "《计算机绘图与图像处理基础》", 31 July 2010 * |
| 李红匣: "基于OpenCV的人眼瞳孔定位算法", 《电子质量》 * |
| 王文成等: "一种基于区域投影的人眼精确定位方法", 《光电子激光》 * |
| 贾永红: "《数字图像处理》", 31 July 2015, 武汉大学出版社 * |
| 顾庆: "基于视觉信息的运动体位置提取算法和实验研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10878593B2 (en) | 2016-12-15 | 2020-12-29 | Tencent Technology (Shenzhen) Company Limited | Pupil localizing method and system |
| WO2018108124A1 (en) * | 2016-12-15 | 2018-06-21 | 腾讯科技（深圳）有限公司 | Method and system for positioning pupil |
| CN108229252A (en) * | 2016-12-15 | 2018-06-29 | 腾讯科技（深圳）有限公司 | A kind of pupil positioning method and system |
| CN107301391B (en) * | 2017-06-16 | 2020-02-07 | 广州市百果园信息技术有限公司 | Area determination method and device and storage medium |
| CN107301391A (en) * | 2017-06-16 | 2017-10-27 | 广州市百果园信息技术有限公司 | Area determination method and device, storage medium |
| CN108498064A (en) * | 2018-03-19 | 2018-09-07 | 中国人民解放军空军航空医学研究所 | Digitize bright spot scintillation detection system |
| CN110415252A (en) * | 2018-04-26 | 2019-11-05 | 北京连心医疗科技有限公司 | A kind of eye circumference organ segmentation method, equipment and storage medium based on CNN |
| CN110415252B (en) * | 2018-04-26 | 2022-08-05 | 北京连心医疗科技有限公司 | CNN-based periocular organ segmentation method, CNN-based periocular organ segmentation equipment and CNN-based periocular organ segmentation storage medium |
| CN108921010A (en) * | 2018-05-15 | 2018-11-30 | 北京环境特性研究所 | A kind of pupil detection method and detection device |
| CN109190509A (en) * | 2018-08-13 | 2019-01-11 | 阿里巴巴集团控股有限公司 | A kind of personal identification method, device and computer readable storage medium |
| US10776646B2 (en) | 2018-08-13 | 2020-09-15 | Alibaba Group Holding Limited | Identification method and apparatus and computer-readable storage medium |
| US11126878B2 (en) | 2018-08-13 | 2021-09-21 | Advanced New Technologies Co., Ltd. | Identification method and apparatus and computer-readable storage medium |
| CN110598635A (en) * | 2019-09-12 | 2019-12-20 | 北京大学第一医院 | Method and system for face detection and pupil positioning in continuous video frames |
| CN110598635B (en) * | 2019-09-12 | 2022-05-27 | 北京大学第一医院 | Method and system for face detection and pupil positioning in continuous video frames |
| CN112733570A (en) * | 2019-10-14 | 2021-04-30 | 北京眼神智能科技有限公司 | Glasses detection method and device, electronic equipment and storage medium |
| CN111488845A (en) * | 2020-04-16 | 2020-08-04 | 深圳市瑞立视多媒体科技有限公司 | Eye sight detection method, device, equipment and storage medium |
| CN112070806A (en) * | 2020-09-14 | 2020-12-11 | 北京华严互娱科技有限公司 | Real-time pupil tracking method and system based on video image |
| FR3132967A1 (en) | 2022-02-23 | 2023-08-25 | Patrick CHARREYRON | Method for generating a reconstituted image and device for generating such a reconstituted image. |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106203375A (en) | A kind of based on face in facial image with the pupil positioning method of human eye detection | |
| CN104978550B (en) | Face identification method based on extensive face database and system | |
| Gu et al. | Feature points extraction from faces | |
| CN103632136B (en) | Human-eye positioning method and device | |
| CN101350063B (en) | Method and apparatus for locating human face characteristic point | |
| CN101923645B (en) | Iris splitting method suitable for low-quality iris image in complex application context | |
| CN104123543B (en) | A kind of eye movement recognition methods based on recognition of face | |
| CN103218605B (en) | A kind of fast human-eye positioning method based on integral projection and rim detection | |
| CN103942577A (en) | Identity identification method based on self-established sample library and composite characters in video monitoring | |
| CN103902978B (en) | Face datection and recognition methods | |
| US20070154096A1 (en) | Facial feature detection on mobile devices | |
| CN104915656B (en) | A kind of fast human face recognition based on Binocular vision photogrammetry technology | |
| CN103413119A (en) | Single sample face recognition method based on face sparse descriptors | |
| CN107066969A (en) | A kind of face identification method | |
| CN106682641A (en) | Pedestrian identification method based on image with FHOG- LBPH feature | |
| CN103310194A (en) | Method for detecting head and shoulders of pedestrian in video based on overhead pixel gradient direction | |
| CN103810491A (en) | Head posture estimation interest point detection method fusing depth and gray scale image characteristic points | |
| CN103049758A (en) | Method for realizing remote authentication by fusing gait flow images (GFI) and head and shoulder procrustes mean shapes (HS-PMS) | |
| CN103020614B (en) | Based on the human motion identification method that space-time interest points detects | |
| CN105512630B (en) | Human eye detection and localization method | |
| CN104008364A (en) | Face recognition method | |
| CN105426882A (en) | Method for rapidly positioning human eyes in human face image | |
| CN103425985B (en) | A kind of face wrinkles on one's forehead detection method | |
| CN110348289A (en) | A kind of finger vein identification method based on binary map | |
| CN103077383B (en) | Based on the human motion identification method of the Divisional of spatio-temporal gradient feature |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161207 |
|
| RJ01 | Rejection of invention patent application after publication |