CN106971177A - A kind of driver tired driving detection method - Google Patents
A kind of driver tired driving detection method Download PDFInfo
- Publication number
- CN106971177A CN106971177A CN201710328421.0A CN201710328421A CN106971177A CN 106971177 A CN106971177 A CN 106971177A CN 201710328421 A CN201710328421 A CN 201710328421A CN 106971177 A CN106971177 A CN 106971177A
- Authority
- CN
- China
- Prior art keywords
- driver
- eyes
- sample
- tired driving
- detection method
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/59—Context or environment of the image inside of a vehicle, e.g. relating to seat occupancy, driver state or inner lighting conditions
- G06V20/597—Recognising the driver's state or behaviour, e.g. attention or drowsiness
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/469—Contour-based spatial representations, e.g. vector-coding
- G06V10/473—Contour-based spatial representations, e.g. vector-coding using gradient analysis
Abstract
The invention discloses a kind of driver tired driving detection method, comprise the following steps:S1:Driving image is acquired using camera;S2:Driver face is detected using AdaBoost algorithms;S3:Positioning and feature extraction to driver's eyes;S4:The calculating of driver's eyes state;S5:The judgement of driver tired driving.The present invention is detected using AdaBoost algorithms to driver face, obtain the gradient matrix of driver's face area figure vertical direction, and floor projection is carried out to gradient matrix, the relative position of eyes in the picture is obtained by the architectural feature of driver face, eyes opening and closing is determined according to distance.Then the parameter of each state of driver's eyes is obtained according to PERCLOS measuring principles, judges whether driver is in fatigue driving state finally by each index and the relation of given threshold.This method has higher accuracy of detection.
Description
Technical field
Present invention relates particularly to a kind of driver tired driving detection method.
Background technology
With the gradually development of Chinese transportation transport service, the generation of traffic accident is more and more frequent, and driver fatigue is driven
Sail and had changed into the one of the main reasons for causing traffic accident, can be mentioned in the same breath with driving when intoxicated.Therefore, developing one has
Driver's fatigue drives monitoring method to ensureing that the safety of people's trip possesses critical significance, becomes relevant scholar and inquires into
Key subject, obtain more and more widely noting.
During being acquired to driver's driving image, easily there is the driving image with certain gradient,
Traditional recognized based on electroencephalogram is caused to combine the driver tired driving detection method for manipulating feature, due to need to be to angular standard
Difference is accurately measured with zero-speed percentage, causes effectively realize the problem of detecting driver fatigue.
The content of the invention
The technical problem to be solved in the present invention is to provide a kind of driver tired driving detection method.
A kind of driver tired driving detection method, it is characterised in that comprise the following steps:
S1:Driving image is acquired using camera;
S2:Driver face is detected using AdaBoost algorithms;
S3:Positioning and feature extraction to driver's eyes;
S4:The calculating of driver's eyes state;
S5:The judgement of driver tired driving.
Further, AdaBoost algorithm steps are as follows:
1)Assuming that training sample is, wherein,For describing to treat training sample,For describing to treat training sample set;For describing generic,If,, then it is negative
Sample, i.e., it is not driver face;If, then it is positive sample, i.e., it is driver face, n is used for needed for describing
The sample total of training;
2)The initialization of weight vector:, wherein,,For describing to be trained
The probability distribution situation of sample;
3)Iterative cycles:Weights are normalized by following formula:
;
By weak learning algorithm, the training sample after being normalized in sequence to weights is trained, and obtains Weak Classifier:;
In above-mentioned weights, error rates of weak classifiers is calculated in sequence:
;
The Weak Classifier for selecting error rate minimum, add it in strong classifier;
The weights of each sample are updated in sequence by optimum classifier:
;
In above formula, if i-th sample can by Accurate classification,;If i-th of sample is accurate
Classification, then, simultaneously;
4)Assuming that the number of times of whole process circulation is T, then the strong classifier finally obtained can be described as follows:
;
In formula,。
Further, the localization method of driver's eyes is as follows:
1)Obtain the gradient matrix of driver's face area figure vertical direction:
;
2)Floor projection is carried out to gradient matrix:
。
Further, the computational methods of driver's eyes state are as follows:
1)Blink during the duration once blinks, eyes are closed again to the time needed for the process opened from reaching, and its value can
Obtained by following formula:
;
2)PERCLOS is the percentage shared by the closing time of eyes in the unit interval, and the unit interval takes 6s, then had:
;
In formula, N is used to describe the useful frame number of collected image in 6s;P (t) is used for representing the letter that eye opening level is changed over time
Count, then the closing time ratio once blinked is:
。
Further, the decision method of driver tired driving is as follows:
1)If the duration D (t) of eye closing situation is higher than thresholding Th1, Th1=2.5, then it is assumed that driver fatigue;
2)Frequency of wink is higher than thresholding Th2, Th2=0.6, then it is assumed that driver fatigue;
3)If PERCLOS value F is higher than thresholding Th3, Th3=4.5, then it is assumed that driver fatigue.
The beneficial effects of the invention are as follows:
The present invention is detected using AdaBoost algorithms to driver face, obtains driver's face area figure vertical direction
Gradient matrix, and carry out floor projection to gradient matrix, eyes are obtained in the picture by the architectural feature of driver face
Relative position, is determined according to distance to eyes opening and closing.Then according to PERCLOS measuring principles obtain driver's eyes each
The parameter of state, judges whether driver is in fatigue driving state finally by each index and the relation of given threshold.Should
Method has higher accuracy of detection.
Embodiment
The present invention is further elaborated for specific examples below, but not as a limitation of the invention.
A kind of driver tired driving detection method, comprises the following steps:
S1:Driving image is acquired using camera;
S2:Driver face is detected using AdaBoost algorithms;
S3:Positioning and feature extraction to driver's eyes;
S4:The calculating of driver's eyes state;
S5:The judgement of driver tired driving.
AdaBoost algorithm steps are as follows:
1)Assuming that training sample is, wherein,For describing to treat training sample,For describing to treat training sample set;For describing generic,If,, then it is negative
Sample, i.e., it is not driver face;If, then it is positive sample, i.e., it is driver face, n is used for needed for describing
The sample total of training;
2)The initialization of weight vector:, wherein,,For describing to be trained
The probability distribution situation of sample;
3)Iterative cycles:Weights are normalized by following formula:
;
By weak learning algorithm, the training sample after being normalized in sequence to weights is trained, and obtains Weak Classifier:;
In above-mentioned weights, error rates of weak classifiers is calculated in sequence:
;
The Weak Classifier for selecting error rate minimum, add it in strong classifier;
The weights of each sample are updated in sequence by optimum classifier:
;
In above formula, if i-th sample can by Accurate classification,;If i-th of sample is accurate
Classification, then, simultaneously;
4)Assuming that the number of times of whole process circulation is T, then the strong classifier finally obtained can be described as follows:
;
In formula,。
The localization method of driver's eyes is as follows:
1)Obtain the gradient matrix of driver's face area figure vertical direction:
;
2)Floor projection is carried out to gradient matrix:
。
The computational methods of driver's eyes state are as follows:
1)Blink during the duration once blinks, eyes are closed again to the time needed for the process opened from reaching, and its value can
Obtained by following formula:
;
2)PERCLOS is the percentage shared by the closing time of eyes in the unit interval, and the unit interval takes 6s, then had:
;
In formula, N is used to describe the useful frame number of collected image in 6s;P (t) is used for representing the letter that eye opening level is changed over time
Count, then the closing time ratio once blinked is:
。
The decision method of driver tired driving is as follows:
1)If the duration D (t) of eye closing situation is higher than thresholding Th1, Th1=2.5, then it is assumed that driver fatigue;
2)Frequency of wink is higher than thresholding Th2, Th2=0.6, then it is assumed that driver fatigue;
3)If PERCLOS value F is higher than thresholding Th3, Th3=4.5, then it is assumed that driver fatigue.
Claims (5)
1. a kind of driver tired driving detection method, it is characterised in that comprise the following steps:
S1:Driving image is acquired using camera;
S2:Driver face is detected using AdaBoost algorithms;
S3:Positioning and feature extraction to driver's eyes;
S4:The calculating of driver's eyes state;
S5:The judgement of driver tired driving.
2. driver tired driving detection method according to claim 1, it is characterised in that AdaBoost algorithm steps are such as
Under:
1)Assuming that training sample is, wherein,For describing to treat training sample,
For describing to treat training sample set;For describing generic,If,, then it is negative sample, i.e.,
It is not driver face;If, then it is positive sample, i.e., it is driver face, n is used for the sample of training needed for describing
This total amount;
2)The initialization of weight vector:, wherein,,For describing to train sample
This probability distribution situation;
3)Iterative cycles:Weights are normalized by following formula:
;
By weak learning algorithm, the training sample after being normalized in sequence to weights is trained, and obtains Weak Classifier:;
In above-mentioned weights, error rates of weak classifiers is calculated in sequence:
;
The Weak Classifier for selecting error rate minimum, add it in strong classifier;
The weights of each sample are updated in sequence by optimum classifier:
;
In above formula, if i-th sample can by Accurate classification,;If i-th of sample is accurate
Classification, then, simultaneously;
4)Assuming that the number of times of whole process circulation is T, then the strong classifier finally obtained can be described as follows:
;
In formula,。
3. driver tired driving detection method according to claim 1, it is characterised in that the positioning side of driver's eyes
Method is as follows:
1)Obtain the gradient matrix of driver's face area figure vertical direction:
;
2)Floor projection is carried out to gradient matrix:
。
4. driver tired driving detection method according to claim 1, it is characterised in that the meter of driver's eyes state
Calculation method is as follows:
1)Blink during the duration once blinks, eyes are closed again to the time needed for the process opened from reaching, and its value can
Obtained by following formula:
;
2)PERCLOS is the percentage shared by the closing time of eyes in the unit interval, and the unit interval takes 6s, then had:
;
In formula, N is used to describe the useful frame number of collected image in 6s;P (t) is used for representing the letter that eye opening level is changed over time
Count, then the closing time ratio once blinked is:
。
5. driver tired driving detection method according to claim 1, it is characterised in that driver tired driving is sentenced
Determine method as follows:
1)If the duration D (t) of eye closing situation is higher than thresholding Th1, Th1=2.5, then it is assumed that driver fatigue;
2)Frequency of wink is higher than thresholding Th2, Th2=0.6, then it is assumed that driver fatigue;
3)If PERCLOS value F is higher than thresholding Th3, Th3=4.5, then it is assumed that driver fatigue.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710328421.0A CN106971177A (en) | 2017-05-11 | 2017-05-11 | A kind of driver tired driving detection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710328421.0A CN106971177A (en) | 2017-05-11 | 2017-05-11 | A kind of driver tired driving detection method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106971177A true CN106971177A (en) | 2017-07-21 |
Family
ID=59331498
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710328421.0A Withdrawn CN106971177A (en) | 2017-05-11 | 2017-05-11 | A kind of driver tired driving detection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106971177A (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101216887A (en) * | 2008-01-04 | 2008-07-09 | 浙江大学 | An automatic computer authentication method for photographic faces and living faces |
CN101950355A (en) * | 2010-09-08 | 2011-01-19 | 中国人民解放军国防科学技术大学 | Method for detecting fatigue state of driver based on digital video |
CN102054163A (en) * | 2009-10-27 | 2011-05-11 | 南京理工大学 | Method for testing driver fatigue based on monocular vision |
-
2017
- 2017-05-11 CN CN201710328421.0A patent/CN106971177A/en not_active Withdrawn
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101216887A (en) * | 2008-01-04 | 2008-07-09 | 浙江大学 | An automatic computer authentication method for photographic faces and living faces |
CN102054163A (en) * | 2009-10-27 | 2011-05-11 | 南京理工大学 | Method for testing driver fatigue based on monocular vision |
CN101950355A (en) * | 2010-09-08 | 2011-01-19 | 中国人民解放军国防科学技术大学 | Method for detecting fatigue state of driver based on digital video |
Non-Patent Citations (1)
Title |
---|
宫法明: ""交通驾驶员脸疲劳驾驶行为优化图像识"", 《计算机仿真》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN100592322C (en) | An automatic computer authentication method for photographic faces and living faces | |
CN105426870B (en) | A kind of face key independent positioning method and device | |
CN109359536B (en) | Passenger behavior monitoring method based on machine vision | |
CN107491769A (en) | Method for detecting fatigue driving and system based on AdaBoost algorithms | |
CN105260705B (en) | A kind of driver's making and receiving calls behavioral value method suitable under multi-pose | |
CN109840565A (en) | A kind of blink detection method based on eye contour feature point aspect ratio | |
CN102819733B (en) | Rapid detection fuzzy method of face in street view image | |
CN106295522A (en) | A kind of two-stage anti-fraud detection method based on multi-orientation Face and environmental information | |
CN103488993B (en) | A kind of crowd's abnormal behaviour recognition methods based on FAST | |
CN101908140A (en) | Biopsy method for use in human face identification | |
CN106156688A (en) | A kind of dynamic human face recognition methods and system | |
CN102085099B (en) | Method and device for detecting fatigue driving | |
CN109376608A (en) | A kind of human face in-vivo detection method | |
CN107133568B (en) | A kind of speed limit prompt and hypervelocity alarm method based on vehicle-mounted forward sight camera | |
CN104298963B (en) | A kind of multi-pose fatigue monitoring method based on face shape regression model of robust | |
CN103903004A (en) | Method and device for fusing multiple feature weights for face recognition | |
CN104182729B (en) | Pedestrian detection method based on ARM embedded platforms | |
CN103020596A (en) | Method for identifying abnormal human behaviors in power production based on block model | |
CN106503645A (en) | Monocular distance-finding method and system based on Android | |
CN110226913A (en) | A kind of self-service examination machine eyesight detection intelligent processing method and device | |
CN109353907A (en) | The security prompt method and system of elevator operation | |
CN107169437A (en) | The method for detecting fatigue driving of view-based access control model | |
CN111104817A (en) | Fatigue detection method based on deep learning | |
CN110222608A (en) | A kind of self-service examination machine eyesight detection intelligent processing method | |
CN110176128A (en) | A kind of driver tired driving alarming processing system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WW01 | Invention patent application withdrawn after publication | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20170721 |