CN107862298A - It is a kind of based on the biopsy method blinked under infrared eye - Google Patents
It is a kind of based on the biopsy method blinked under infrared eye Download PDFInfo
- Publication number
- CN107862298A CN107862298A CN201711204490.7A CN201711204490A CN107862298A CN 107862298 A CN107862298 A CN 107862298A CN 201711204490 A CN201711204490 A CN 201711204490A CN 107862298 A CN107862298 A CN 107862298A
- Authority
- CN
- China
- Prior art keywords
- max
- blink
- frame
- detected
- doubtful
- 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.)
- Granted
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
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
-
- 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
- G06V40/40—Spoof detection, e.g. liveness detection
- G06V40/45—Detection of the body part being alive
Landscapes
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of based on the biopsy method blinked under infrared camera, belong to digital image processing techniques field.The present invention is：One section of video-frequency band to be detected for meeting detection duration is gathered by infrared eye；Obtain facial image and the right and left eyes position of the present frame of video-frequency band to be detected；The facial image of present frame is divided into two regions up and down, wherein upper region includes right and left eyes position；And extract the frame difference figure in the upper region of present frame and previous frame；After carrying out binary conversion treatment to frame difference figure, carry out morphological image opens operation, then searches all profiles, if number of contours more than two, carries out doubtful blink detection and obtains blink position；Otherwise next frame is continued with；Again to carrying out true blink detection to present frame based on blink position：Then think that current object to be detected is live body in the blink position of left and right respectively in left and right eye position.Invention increases the robustness of blink detection so that false drop rate significantly reduces.
Description
Technical field
The invention belongs to digital image processing techniques field, and in particular to a kind of based on the work blinked under infrared eye
Body detecting method.
Background technology
With developing rapidly for Computer Applied Technology and artificial intelligence disciplinary study so that there is an urgent need to should in many fields
With image processing techniques, the application field of digital image processing techniques is more and more, wherein being achieved in face identification system
Utilization and effect well.Face recognition technology is in identity recognizing technology, and due to its convenience, prospect is seen by people always
It is good, but requirement of the identity recognizing technology for security is very high, traditional single recognition of face can only ensure to verify face
Accuracy, the source for not ensuring that face is real people, or the feature that photo, video etc. is counterfeit, so as to cause people
Face identifies that the application in safety-security area is had a greatly reduced quality.
Vivo identification technology is the technology for making a distinction photo, the counterfeit and real face of video, current vivo identification
Technology mainly has following methods：
1) the living body faces detection based on microtexture, has been lacked and poor because the face of photo and video returns in detail
Different, these differences can cause the difference of tiny texture.The detection of living body faces is carried out by textural characteristics.But the method is restricted
In the limitation of scene, fail have good performance under complex illumination and scene.
2) be based on multispectral living body faces detection method, due to true man skin and forge face material in material
Have differences, result in the difference of the refractive index in optical imagery.By being gathered to different-waveband near infrared light for face, come
Carry out the judgement of live body.Good effect is achieved, however it is more harsh for acquisition condition requirement, and equipment cost is relatively
It is high.
3) the living body faces detection method based on movable information, this method rely primarily on the motion of face specific region, such as
Blink, shake the head and the change of the micro- expression of face etc..Such method plays preferable effect for the counterfeit of photo, but can not
Effectively resisted for counterfeit make of video.
The content of the invention
The goal of the invention of the present invention is：For above-mentioned problem, there is provided one kind can be same under infrared eye
When prevent In vivo detection technology that photo and video are attacked.
The present invention based on the biopsy method blinked under infrared eye, comprise the following steps：
One section of video-frequency band to be detected for meeting detection duration is gathered by infrared eye；
Obtain facial image and the right and left eyes position of the present frame of video-frequency band to be detected：Image is carried out to current frame image to turn
Processing is changed, obtains frame gray-scale map；And Face datection is carried out to frame gray-scale map, obtain facial image；Pass through facial feature points detection
Method obtains the right and left eyes position of facial image；
Doubtful blink detection is carried out to the present frame of video-frequency band to be detected：
The facial image of present frame is divided into two regions up and down, wherein upper region includes right and left eyes position；And extract
The frame difference figure in the upper region of present frame and previous frame；
After carrying out binary conversion treatment to frame difference figure, carry out morphological image opens operation, then searches all profiles, if profile
Number is more than 2, then carries out doubtful blink detection, determine doubtful blink position；Otherwise next frame is continued with；
Doubtful blink detection is：The maximum profile of contour area the first two is searched, is designated as Smax_1、Smax_2, wherein Smax_1's
Contour area is more than Smax_2If meet four blink conditions simultaneously, then it is assumed that doubtful blink be present, and by Smax_1、Smax_2As
Doubtful blink position；
Four blink conditions are：
(1)Smax_1、Smax_2Area it is close：Smax_1Contour area be not more than Smax_2150%.
(2) area of other profiles is less than Smax_250%.
(3)Smax_1、Smax_2Line and horizontal angle between midpoint are no more than 30 degree.
(4)Smax_1、Smax_2The distance between midpoint is no more than k times of frontier distance D, wherein frontier distance D signs Smax_1's
Maximum boundary distance, k value is 1.5-3；
Based on position coordinates, to the profile S for detecting to obtain in step 3.3max_1、Smax_2Distinguish left and right profile；
True blink detection is carried out to the present frame of video-frequency band to be detected：
Based on position coordinates, to doubtful blink position Smax_1、Smax_2Distinguish left and right blink position；
Judge whether left and right eye position blinks in position in left and right respectively；If it is not, then continue with next frame；If so,
Then think that current object to be detected is live body.
In summary, by adopting the above-described technical solution, the beneficial effects of the invention are as follows：
(1) facial feature points detection and frame difference method are combined, frame difference method is the detection of doubtful blink action, Ran Houtong
The characteristic point that detects is crossed to verify whether being blink, to determine whether the access of true people, adds the Shandong of blink detection
Rod so that false drop rate significantly reduces.
(2) infrared eye of additional source of light has been used to obtain image, the infrared eye of additional infrared light source can
To prevent from getting the image of video equipment, avoid counterfeiter from acting deception by the blink in video and passed through checking.
Embodiment
To make the object, technical solutions and advantages of the present invention clearer, with reference to embodiment, the present invention is made into
One step it is described in detail.
It is a kind of based on the biopsy method blinked under infrared eye, include following step：
Step 1：Gather face video data.
One section is gathered on to be detected by infrared eye (camera device of additional infrared light source and optical filter)
Video data, the input image data as the present invention；
Step 2：Facial image and position of human eye detection.
First by two field picture (the original image I of input) carry out gradation conversion, obtain gray level image I_gray, afterwards for
Image I_gray carries out being Face datection acquisition face rectangle frame RECT by Viola-Jones human-face detectors, obtains current
The facial image (i.e. image-region determined by the face rectangle frame) of frame.Wherein Viola-Jones graders are first by class
Haar input feature vectors carry out for image-region and/or poor progress binaryzation, which use integral image techniques speed-up computation,
Then the grader node of Face datection is created using adaboost, finally the composition screening type cascade of grader node, group
Into a complete human-face detector.
Then human face characteristic point extraction is carried out again.The I_gray and facial image obtained according to upper step, in facial image
Enterprising pedestrian's face characteristic point extraction, obtain the position P1, P2 of two pupils of human eye.In the present embodiment, pass through ASM (Active
Shape Model) algorithm does feature point extraction, and 68 characteristic points of face are obtained, every 6 characteristic point positions be averaged, acquisition
Position P1, the P2 of two pupils, wherein P1, P2 correspond to left and right pupil respectively.
The present embodiment only provides one preferably, specific to implement combination, and this method remains to acquisition one when eyes are closed
Individual accurate position of human eye, the position of right and left eyes can also be obtained with the such as direct human eye detection of other means.
Step 3：Search doubtful blink position.
Step 3.1：Facial image is divided into two regions up and down, wherein upper region includes human eye, i.e. two pupils
Position P1, P2.And just the gray-scale map in region, the frame difference for obtaining present frame and previous frame scheme I_di thereon.
Step 3.2：Image binaryzation processing carried out using pixel value 5 as threshold value to frame difference figure I_di, acquisition binaryzation it
Rear image I_bi simultaneously opens image I_bi operation, eliminates noise.Wherein open operation for morphological image from processing, that is, pass through
The operation that structural element B is first carried out corroding to set A and then expanded again with B to result.Typically make object in bianry image
Profile become smooth, interruption narrow in bianry image can be disconnected and eliminate thin protrusion.
Step 3.3：All profiles in binary image I_bi are searched, if number of contours is more than 2, doubtful blink be present,
And the detection of doubtful blink position is carried out, otherwise continue with next frame；
The detection of doubtful blink position：In the case where number of contours is more than two, the maximum wheel of contour area the first two is searched
Exterior feature, it is designated as Smax_1、Smax_2, wherein Smax_1＞ Smax_2When descending condition while meeting, it is believed that doubtful blink situation occurs：
(1)Smax_1、Smax_2Area it is close：Smax_1Contour area be not more than Smax_2150%.
(2) area of other profiles is less than Smax_250%.
(3)Smax_1、Smax_2Line and horizontal angle between midpoint are no more than 30 degree.
(4)Smax_1、Smax_2The distance between midpoint is no more than k times of frontier distance D, wherein frontier distance D signs Smax_1's
Maximum boundary distance, k value is 1.5-3.
Step 4：Authenticity of blinking detection, the i.e. In vivo detection based on blink detection.
Based on position coordinates, to the profile S for detecting to obtain in step 3.3max_1、Smax_2Distinguish left and right profile；
And judge left and right eye position whether respectively in the profile of left and right, i.e., whether left pupil P1 is and right in left profile
Whether pupil P2 is in right wheel exterior feature, if so, then identification is counter has given birth to true blink；If it is not, then detection period (being preferably 10 seconds)
Interior repeat step 2-4, if detecting true blink within the detection period, then it is assumed that current object to be detected is live body, passes through inspection
Survey；If it is not detected by true blink within the detection period, then it is assumed that to be counterfeit, testing result is non-living body.
The foregoing is only a specific embodiment of the invention, any feature disclosed in this specification, except non-specifically
Narration, can alternative features equivalent by other or with similar purpose replaced；Disclosed all features or all sides
Method or during the step of, in addition to mutually exclusive feature and/or step, can be combined in any way.
Claims (4)
- It is 1. a kind of based on the biopsy method blinked under infrared eye, it is characterised in that to comprise the following steps：One section of video-frequency band to be detected for meeting detection duration is gathered by infrared eye；Obtain facial image and the right and left eyes position of the present frame of video-frequency band to be detected：Current frame image is carried out at image conversion Reason, obtains frame gray-scale map；And Face datection is carried out to frame gray-scale map, obtain facial image；Obtained by facial feature points detection method Take the right and left eyes position of facial image；Doubtful blink detection is carried out to the present frame of video-frequency band to be detected：The facial image of present frame is divided into two regions up and down, wherein upper region includes right and left eyes position；And extract current The frame difference figure in the upper region of frame and previous frame；After carrying out binary conversion treatment to frame difference figure, carry out morphological image opens operation, then searches all profiles, if number of contours is big In 2, then doubtful blink detection is carried out, determines doubtful blink position；Otherwise next frame is continued with；Doubtful blink detection is：The maximum profile of contour area the first two is searched, is designated as Smax_1、Smax_2, wherein Smax_1Profile Area is more than Smax_2If meet four blink conditions simultaneously, then it is assumed that doubtful blink be present, and by Smax_1、Smax_2As doubtful Blink position；Four blink conditions are：(1)Smax_1、Smax_2Area it is close：Smax_1Contour area be not more than Smax_2150%；(2) area of other profiles is less than Smax_250%；(3)Smax_1、Smax_2Line and horizontal angle between midpoint are no more than 30 degree；(4)Smax_1、Smax_2The distance between midpoint is no more than k times of frontier distance D, wherein frontier distance D signs Smax_1Maximum Frontier distance, k value is 1.5-3；Based on position coordinates, to the profile S for detecting to obtain in step 3.3max_1、Smax_2Distinguish left and right profile；True blink detection is carried out to the present frame of video-frequency band to be detected：Based on position coordinates, to doubtful blink position Smax_1、Smax_2Distinguish left and right blink position；Judge whether left and right eye position blinks in position in left and right respectively；If it is not, then continue with next frame；If so, then recognize It is live body for current object to be detected.
- 2. as claimed in claim 1 based on the biopsy method blinked under infrared eye, it is characterised in that obtain The facial image for taking the present frame of video-frequency band to be detected is specially：Face datection is done to frame gray-scale map by Viola-Jones human-face detectors and obtains face rectangle frame, obtains present frame Facial image.
- 3. as claimed in claim 2 based on the biopsy method blinked under infrared eye, it is characterised in that obtain people The right and left eyes position of face image is：Feature point extraction is done to facial image by Active Shape Model algorithms, 68 characteristic points of face are obtained, by every 6 Individual characteristic point position is averaged, and extracts the position of two pupils of human eye.
- 4. as claimed in claim 1 based on the biopsy method blinked under infrared eye, it is characterised in that described to treat Detect video-frequency band when a length of 10 seconds.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711204490.7A CN107862298B (en) | 2017-11-27 | 2017-11-27 | Winking living body detection method based on infrared camera device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711204490.7A CN107862298B (en) | 2017-11-27 | 2017-11-27 | Winking living body detection method based on infrared camera device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107862298A true CN107862298A (en) | 2018-03-30 |
| CN107862298B CN107862298B (en) | 2021-07-06 |
Family
ID=61702847
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711204490.7A Active CN107862298B (en) | 2017-11-27 | 2017-11-27 | Winking living body detection method based on infrared camera device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107862298B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108549887A (en) * | 2018-07-23 | 2018-09-18 | 北京智芯原动科技有限公司 | A kind of living body faces detection method and device |
| CN109522868A (en) * | 2018-11-30 | 2019-03-26 | 北京七鑫易维信息技术有限公司 | A kind of method and apparatus of detection blink |
| CN110059624A (en) * | 2019-04-18 | 2019-07-26 | 北京字节跳动网络技术有限公司 | Method and apparatus for detecting living body |
| CN110738072A (en) * | 2018-07-18 | 2020-01-31 | 浙江宇视科技有限公司 | Living body judgment method and device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060188130A1 (en) * | 2005-01-20 | 2006-08-24 | Samsung Electronics Co., Ltd. | Apparatus and method for normalizing face image used for detecting drowsy driving |
| CN104200146A (en) * | 2014-08-29 | 2014-12-10 | 华侨大学 | Identity verifying method with video human face and digital lip movement password combined |
| CN107239735A (en) * | 2017-04-24 | 2017-10-10 | 复旦大学 | A kind of biopsy method and system based on video analysis |
| CN107330378A (en) * | 2017-06-09 | 2017-11-07 | 湖北天业云商网络科技有限公司 | A kind of driving behavior detecting system based on embedded image processing |
| CN107346422A (en) * | 2017-06-30 | 2017-11-14 | 成都大学 | A kind of living body faces recognition methods based on blink detection |
-
2017
- 2017-11-27 CN CN201711204490.7A patent/CN107862298B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060188130A1 (en) * | 2005-01-20 | 2006-08-24 | Samsung Electronics Co., Ltd. | Apparatus and method for normalizing face image used for detecting drowsy driving |
| CN104200146A (en) * | 2014-08-29 | 2014-12-10 | 华侨大学 | Identity verifying method with video human face and digital lip movement password combined |
| CN107239735A (en) * | 2017-04-24 | 2017-10-10 | 复旦大学 | A kind of biopsy method and system based on video analysis |
| CN107330378A (en) * | 2017-06-09 | 2017-11-07 | 湖北天业云商网络科技有限公司 | A kind of driving behavior detecting system based on embedded image processing |
| CN107346422A (en) * | 2017-06-30 | 2017-11-14 | 成都大学 | A kind of living body faces recognition methods based on blink detection |
Non-Patent Citations (4)
| Title |
|---|
| JIAN WANG 等: "Eye detection algorithm base on area-blocks pairing and multi-angle template matching", 《2009 CHINESE CONTROL AND DECISION CONFERENCE》 * |
| LITING WANG 等: "Eye Blink Detection Based on Eye Contour Extraction", 《THE INTERNATIONAL SOCIETY FOR OPTICAL ENGINEERING》 * |
| 宫文策 等: "差别光照下基于人眼轮廓的疲劳值算法研究", 《信息与电脑（理论版）》 * |
| 林诚燕: "基于人眼序列分析的活体人脸检测方法", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110738072A (en) * | 2018-07-18 | 2020-01-31 | 浙江宇视科技有限公司 | Living body judgment method and device |
| CN108549887A (en) * | 2018-07-23 | 2018-09-18 | 北京智芯原动科技有限公司 | A kind of living body faces detection method and device |
| CN109522868A (en) * | 2018-11-30 | 2019-03-26 | 北京七鑫易维信息技术有限公司 | A kind of method and apparatus of detection blink |
| CN109522868B (en) * | 2018-11-30 | 2021-07-23 | 北京七鑫易维信息技术有限公司 | Method and device for detecting blink |
| CN110059624A (en) * | 2019-04-18 | 2019-07-26 | 北京字节跳动网络技术有限公司 | Method and apparatus for detecting living body |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107862298B (en) | 2021-07-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Learning deep models for face anti-spoofing: Binary or auxiliary supervision | |
| Davison et al. | Objective micro-facial movement detection using facs-based regions and baseline evaluation | |
| Ghimire et al. | A robust face detection method based on skin color and edges | |
| CN109558810B (en) | Target person identification method based on part segmentation and fusion | |
| US6661907B2 (en) | Face detection in digital images | |
| CN103942577B (en) | Based on the personal identification method for establishing sample database and composite character certainly in video monitoring | |
| CN106557723B (en) | Face identity authentication system with interactive living body detection and method thereof | |
| CN108921041A (en) | A kind of biopsy method and device based on RGB and IR binocular camera | |
| CN106682578B (en) | Weak light face recognition method based on blink detection | |
| CN104361326A (en) | Method for distinguishing living human face | |
| CN105138954A (en) | Image automatic screening, query and identification system | |
| CN107862298A (en) | It is a kind of based on the biopsy method blinked under infrared eye | |
| CN110008793A (en) | Face identification method, device and equipment | |
| Mehta et al. | Motion and region aware adversarial learning for fall detection with thermal imaging | |
| CN108446690A (en) | A kind of human face in-vivo detection method based on various visual angles behavioral characteristics | |
| Lopar et al. | An overview and evaluation of various face and eyes detection algorithms for driver fatigue monitoring systems | |
| Yuan et al. | Ear detection based on CenterNet | |
| Guan | Robust eye detection from facial image based on multi-cue facial information | |
| KR101711328B1 (en) | Method for classifying children and adult by using head and body height in images obtained from camera such as CCTV | |
| CN110765925A (en) | Carrier detection and gait recognition method based on improved twin neural network | |
| CN112613430B (en) | Gait recognition method based on deep migration learning | |
| De Siqueira et al. | Adaptive detection of human skin in color images | |
| Chakraborty | An illumination invariant face detection based on human shape analysis and skin color information | |
| Li et al. | The face localization and regional features extraction | |
| KR20160017152A (en) | Gender Classification Based on Binary Haar Cascade |
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 |