CN105235451B - Driver's vision identifying system with tire pressure monitoring function - Google Patents

Abstract

The invention discloses a kind of driver's vision identifying system with tire pressure monitoring function, detector is installed on wheel, when the air pressure of certain tire is reduced, the weight of vehicle can make the rolling radius of the wheel diminish, cause its rotating ratio other wheels fast, by the rotating speed difference between comparative tire, to monitor tire pressure, wheel speed sensors are determining wheel whether locking, so as to decide whether to start anti-lock braking system, when tire pressure is reduced, the weight of vehicle can make diameter of tyres diminish, this may result in speed and changes, this change gives a warning to driver i.e. for triggering the warning system in visual identifying system；The method for expressing and front benchmark image of head pose, Image Feature Matching, head pose estimation, experimental configuration analysis.The system also can be detected to the vision of driver during to automotive tyre pressure monitoring constantly so that driver's vision in driving procedure is in normal range all the time, it is to avoid cause security incident to occur.

Description

Driver's vision identifying system with tire pressure monitoring function

Technical field

The present invention relates to automotive field, in particular it relates to a kind of identification of the driver's vision with tire pressure monitoring function system System.

Background technology

Automobile is defined as below：The vehicle carried by power drive, the non-track with 4 or more than 4 wheels, mainly For：Carrying personnel and/or goods；Traction carrying personnel or the vehicle of goods；Specific use.1879, German engineer card Er Benci, tests the tentative electromotor of two-stroke first successfully.In October, 1883, he has founded " Ben Ci companies and Lay Mattress gas motor factory ", 1885, he made first this thatch patent motor vehicles in Mannheim, and the car is three-wheeled motor car, is adopted With 0.9 horsepower of gasoline engine of a two-stroke single cylinder, this car possesses some basic characteristics of Hyundai Motor, such as spark ignition, Water-cooling circulating, steel pipe vehicle frame, Leaf Spring Suspension, rear wheel drive front-wheel steer and binding handle etc..On the January of 1886 29, German engineer karr this thatch patent that has been its motor vehicles application.November in the same year, the three-wheeled motor vehicle of karr this thatch obtain Germany patent right.Here it is first Hyundai Motor in the world generally acknowledged.For above-mentioned reasons, people are typically 1886 Year as the automobile first year, also some scholars make karr this thatch year i.e. 1885 year on first three-wheeled motor car, are considered as automobile It is born year.It is within 1885 1 year of the decisive breakthrough of automobile invention acquirement.Also existed in the Ben Ci of same factory with Daimler at that time Research automobile.He has almost made petrol engine with Daimler simultaneously in 1885, is mounted on automobile, with 12 kilometers per hour Speed traveling, succeed.This year, the Butler of Britain have also been invented gasoline-powered automobile.Additionally, meaning is big The Claude Bernard of profit have also been invented automobile, and the general strange love and two Crinis Carbonisatus of Fu Luobofu of Russia understand the automobile equipped with internal combustion engine.With Front China does not have auto manufacturing.On Chinese soil, first automobile is that the U.S. being input into for 1903 produces the little vapour of oldsmobile board Car, neck number one running car licence, its owner be Shanghai rich man.Start building emerging from July nineteen fifty-three First Automobile Works Build, in July, 1956 goes into operation, and July 13 nineteen fifty-seven, China produced the Jiefang brand automobile of first loading, and in May, 1958, China's First Automobile Works voluntarily Development and design produce first rise and fall with political wind and cloud at that time jolt, share weal or woe it is red Flag board passenger car, is described as " east romantic charm ".In decades, Domestic Automotive Industry has obtained quick development.Particularly reform Since opening, automobile production employs the safety and amenities of various high-tech and hommization, draws the essence of Foreign Automobile scientific research China.Not only grasp and taken advantage of traditional firm moulding, more the gentle and lovely style and features of fashion automobile, the lines flow smoothly, drive comfortable " car " New lover is constantly born.At the bottom of calendar year 2001, China has become a full member of the World Trade Organization, and taking this as an opportunity, and Chinese Auto Industry has welcome one The individual new high-speed developing period.2009, Chinese automobile production and marketing was respectively 1379.10 ten thousand and 1364.48 ten thousand, surmounts at one stroke The U.S., referred to as the first in the world automobile production and marketing big country.The whole year production and marketing of Chinese automobile in 2012 be respectively 1927.18 ten thousand and 1930.64 ten thousand, continue to hold a post or title the first in the world within continuous 4 years.Enter after 10 years high speed developments, the riding driving skills of Chinese independent brand Art has obtained significant progress.The first half of the year in 2013, successively a vapour red flag of listing, gentry of Beijing Automobile Workshop be precious, Chang'an it is farsighted gallop, lucky Deidro Deluxe, BYD think it is sharp and before this successively the upper vapour Roewe of listing, Guangzhou Automobile Workshop pass the high-end passenger car of independent brand for representative such as auspicious to Joint brand is initiated group type and is charged, and is progressively rewritten the present situation that independent brand passenger car can only seize market in low and middle-end.It is existing It is various for kinds of automobile, people driving traveling during, due to driver fatigue or other factors cause vision not Can concentrate, cause unexpected generation.

The content of the invention

The technical problem to be solved is to provide a kind of identification of the driver's vision with tire pressure monitoring function system System, the system also can be detected to the vision of driver during to automotive tyre pressure monitoring constantly so that driver exists In driving procedure, vision is, in normal range, to prevent which from fatigue or other scatterbrained phenomenons occur all the time, So that car steering is safer, it is to avoid cause security incident to occur.

The present invention the adopted technical scheme that solves the above problems is：The identification of driver's vision with tire pressure monitoring function System：

(1) detector is installed on wheel, when the air pressure of certain tire is reduced, the weight of vehicle can make the rolling half of the wheel Footpath will diminish, and cause its rotating ratio other wheels fast, by the rotating speed difference between comparative tire, to reach the mesh of monitoring tire pressure , ABS determines wheel whether locking by wheel speed sensors, so as to decide whether to start anti-lock braking system, works as tire pressure During reduction, the weight of vehicle can make diameter of tyres diminish, and this may result in speed and changes, and this change can be used for triggering Warning system in visual identifying system is giving a warning to driver；

(2) method for expressing of head pose and front benchmark image：The rotation of head pose includes three degree of freedom, that is, divide It is not the rotation around X, Y, Z axis, corresponding motion is referred to as horizontal turn, inclines and pitching, head pose estimation is all relative driving The person of sailing for head pose, is called front benchmark image this attitude during normal driving, using actively setting side When formula, i.e. driver use the system for the first time, posture of driving is kept, is eyed to the front, using image now as front reference map Picture, and using coordinate system now as conventional coordinates；

(3) Image Feature Matching：The video sequence of given input, detects the position of front face, first with front face On the basis of, the attitude angle of labelling front face image is 0 °, and the driver that IP Camera is obtained is regarded in driving procedure Frequency sequence with benchmark image in different angles, light, block and under factor, have very big difference, Image Feature Matching is detected The number of match point directly affects the accuracy of matching result, for the characteristic point for meeting traditional implicit assumption is entered by existing algorithm Row matching, for the region for being unsatisfactory for assumed condition, that is, the region for being difficult to direct matching is matched by the method for optimizing, is come More match points are obtained, using SIFT algorithms to feature point detection and matching are carried out between adjacent two field picture, then using driving The person's of sailing colour of skin priori, is filtered and is tracked to characteristic matching result, obtains the human face region between different images, experiment As a result show, on the image of 70 × 80 pixels, be obtained in that 150 characteristic points；

(4) head pose estimation：Based on the characteristic matching result in human face region, two are estimated using the method for Epipolar geometry A certain characteristic point X in attitude angle between width image, i.e. three dimensions, is projected in two different visual angles, imaging point Position x1, x2, and imaging point is only relevant with the relative attitude between the parameter of video camera and camera, this paper driver head's attitudes, i.e., Consider the relative motion between head and photographic head, then meet：

x^T ₂Fx₁=0

F is fundamental matrix, is the Algebraic Expression of limit restraint；Arrange e1, e2 is antipodal points, i.e., the baseline of two video cameras with The intersection point of imaging plane；C1, C2 are the center of two cameras；1 ' represents the x1 points in image, corresponding in other image To polar curve, i.e.,

1 '=Fx₁

Basis matrix is obtained according to normalized linear 8 points of algorithms first, driver head's attitude is carried out on this basis Angle, if the intrinsic parameter of video camera is K, then camera matrix is：

P=K [P/t]

X=PX is a bit on image, then there is K^-1So that,

Then pictures of 1 point of the space X under video camera normalization matrix isBasis corresponding with normalization camera matrix Matrix for video camera essential matrix with the relation of fundamental matrix is：

E=K^TFK

Essential matrix contains rotation and translates Vector Message, i.e.,：

E=[t]_xR

Spin matrix R can be obtained from essential matrix by above formula, spin matrix can be converted into Eulerian angles, i.e., rotating around Z, The anglec of rotation of Y, X is α, and beta, gamma, spin matrix can be expressed as：

Based on the characteristic matching result in human face region, the basic square between image is estimated using dynamic Attitude estimation algorithm Battle array and essential matrix, and then estimate spaced winding rotary shaft X of any two width adjacent image frame, the relative angle in tri- directions of Y, Z；

(5) experimental configuration analysis：During for carrying out head pose estimation using the system, obtain after being got on the bus using driver Benchmark image be standard front face face coordinate system, if the center of driver's inner eye corner line be zero, taken the photograph by network As each two field picture that head is obtained, enough characteristic points are obtained by Image Feature Matching algorithm, will using RANSAC algorithms The match point of mistake is filtered, and then using Algorithm of Head Pose Estimation, calculates rotation of driver's face relative to conventional coordinates Torque battle array, and then the α of driver head's attitude relative to frontal pose of the frame is estimated, the beta, gamma anglec of rotation works as driver Head roll attitude, i.e., now driver eyes to the front the most of the time, and head has left-right rotation by a small margin, i.e., horizontal Rolling, observes road conditions, and the driver of record is in fatigue state, in driving procedure, head occurs at set intervals significantly Lowly, that is, overlook, then can lift rapidly, and the frequency more and more higher significantly nodded, it is left in driver head's long-time Partially, that is, go off course, persistent period length crosses 120 frames, represent that driver is absent-minded or sleepy, occur overlooking and going off course situation, system can be sent out Go out alarm, remind driver.

After algorithm is calculated, Projection Display visually facilitates driver quickly obtain information, is easy to vehicle The identification of tire pressure, front pedestrian, vehicle and route.

To sum up, the invention has the beneficial effects as follows：The system also can constantly to driving during to automotive tyre pressure monitoring The vision of person is detected so that driver's vision in driving procedure is, in normal range, to prevent its appearance tired all the time Labor or other scatterbrained phenomenons so that car steering is safer, it is to avoid cause security incident to occur.

Specific embodiment

With reference to embodiment, make detailed description further to the present invention, but embodiments of the present invention are not limited to This.

Embodiment：

Driver's vision identifying system with tire pressure monitoring function：

(1) detector is installed on wheel, when the air pressure of certain tire is reduced, the weight of vehicle can make the rolling half of the wheel Footpath will diminish, and cause its rotating ratio other wheels fast, by the rotating speed difference between comparative tire, to reach the mesh of monitoring tire pressure , ABS determines wheel whether locking by wheel speed sensors, so as to decide whether to start anti-lock braking system, works as tire pressure During reduction, the weight of vehicle can make diameter of tyres diminish, and this may result in speed and changes, and this change can be used for triggering Warning system in visual identifying system is giving a warning to driver；

(2) method for expressing of head pose and front benchmark image：The rotation of head pose includes three degree of freedom, that is, divide It is not the rotation around X, Y, Z axis, corresponding motion is referred to as horizontal turn, inclines and pitching, head pose estimation is all relative driving The person of sailing for head pose, is called front benchmark image this attitude during normal driving, using actively setting side When formula, i.e. driver use the system for the first time, posture of driving is kept, is eyed to the front, using image now as front reference map Picture, and using coordinate system now as conventional coordinates；

(3) Image Feature Matching：The video sequence of given input, detects the position of front face, first with front face On the basis of, the attitude angle of labelling front face image is 0 °, and the driver that IP Camera is obtained is regarded in driving procedure Frequency sequence with benchmark image in different angles, light, block and under factor, have very big difference, Image Feature Matching is detected The number of match point directly affects the accuracy of matching result, for the characteristic point for meeting traditional implicit assumption is entered by existing algorithm Row matching, for the region for being unsatisfactory for assumed condition, that is, the region for being difficult to direct matching is matched by the method for optimizing, is come More match points are obtained, using SIFT algorithms to feature point detection and matching are carried out between adjacent two field picture, then using driving The person's of sailing colour of skin priori, is filtered and is tracked to characteristic matching result, obtains the human face region between different images, experiment As a result show, on the image of 70 × 80 pixels, be obtained in that 150 characteristic points；

(4) head pose estimation：Based on the characteristic matching result in human face region, two are estimated using the method for Epipolar geometry A certain characteristic point X in attitude angle between width image, i.e. three dimensions, is projected in two different visual angles, imaging point Position x1, x2, and imaging point is only relevant with the relative attitude between the parameter of video camera and camera, this paper driver head's attitudes, i.e., Consider the relative motion between head and photographic head, then meet：

x^T ₂Fx₁=0

F is fundamental matrix, is the Algebraic Expression of limit restraint；Arrange e1, e2 is antipodal points, i.e., the baseline of two video cameras with The intersection point of imaging plane；C1, C2 are the center of two cameras；1 ' represents the x1 points in image, corresponding in other image To polar curve, i.e.,

1 '=Fx₁

Basis matrix is obtained according to normalized linear 8 points of algorithms first, driver head's attitude is carried out on this basis Angle, if the intrinsic parameter of video camera is K, then camera matrix is：

P=K [P/t]

X=PX is a bit on image, then there is K^-1So that,

Then pictures of 1 point of the space X under video camera normalization matrix isBasis corresponding with normalization camera matrix Matrix for video camera essential matrix with the relation of fundamental matrix is：

E=K^TFK

Essential matrix contains rotation and translates Vector Message, i.e.,：

E=[t]_xR

Spin matrix R can be obtained from essential matrix by above formula, spin matrix can be converted into Eulerian angles, i.e., rotating around Z, The anglec of rotation of Y, X is α, and beta, gamma, spin matrix can be expressed as：

Based on the characteristic matching result in human face region, the basic square between image is estimated using dynamic Attitude estimation algorithm Battle array and essential matrix, and then estimate spaced winding rotary shaft X of any two width adjacent image frame, the relative angle in tri- directions of Y, Z；

(5) experimental configuration analysis：During for carrying out head pose estimation using the system, obtain after being got on the bus using driver Benchmark image be standard front face face coordinate system, if the center of driver's inner eye corner line be zero, taken the photograph by network As each two field picture that head is obtained, enough characteristic points are obtained by Image Feature Matching algorithm, will using RANSAC algorithms The match point of mistake is filtered, and then using Algorithm of Head Pose Estimation, calculates rotation of driver's face relative to conventional coordinates Torque battle array, and then the α of driver head's attitude relative to frontal pose of the frame is estimated, the beta, gamma anglec of rotation works as driver Head roll attitude, i.e., now driver eyes to the front the most of the time, and head has left-right rotation by a small margin, i.e., horizontal Rolling, observes road conditions, and the driver of record is in fatigue state, in driving procedure, head occurs at set intervals significantly Lowly, that is, overlook, then can lift rapidly, and the frequency more and more higher significantly nodded, it is left in driver head's long-time Partially, that is, go off course, persistent period length crosses 120 frames, represent that driver is absent-minded or sleepy, occur overlooking and going off course situation, system can be sent out Go out alarm, remind driver.

The system also can be detected to the vision of driver during to automotive tyre pressure monitoring constantly so that driven Person's vision in driving procedure is, in normal range, to prevent which from fatigue occur or other are scatterbrained existing all the time As so that car steering is safer, it is to avoid cause security incident to occur.

The above, is only presently preferred embodiments of the present invention, not does any pro forma restriction to the present invention, it is every according to According to the present invention technology, method substantially above example is made any simple modification, equivalent variations, each fall within the present invention Protection domain within.

Claims (1)

1. there is the driver's vision identifying system of tire pressure monitoring function, it is characterised in that：

(1) detector is installed on wheel, when the air pressure of certain tire is reduced, the weight of vehicle incites somebody to action can the rolling radius of the wheel Diminish, cause its rotating ratio other wheels fast, by the rotating speed difference between comparative tire, to reach the purpose of monitoring tire pressure, ABS determines wheel whether locking by wheel speed sensors, so as to decide whether to start anti-lock braking system, when tire pressure is reduced When, the weight of vehicle can make diameter of tyres diminish, and this may result in speed and changes, and this change is used for triggering vision knowledge Warning system in other system is giving a warning to driver；

(2) method for expressing of head pose and front benchmark image：The rotation of head pose includes three degree of freedom, that is, be respectively Around the rotation of X, Y, Z axis, corresponding motion is referred to as horizontal turn, inclines and pitching, and head pose estimation is all relative driver During normal driving for head pose, this attitude is called front benchmark image, using active set-up mode, i.e., When driver uses the system for the first time, posture of driving is kept, is eyed to the front, using image now as front benchmark image, and Using coordinate system now as conventional coordinates；

(3) Image Feature Matching：The video sequence of given input, detects the position of front face, first with front face as base Standard, the attitude angle of labelling front face image is 0 °, and the driver that IP Camera is obtained video sequence in driving procedure Row with benchmark image in different angles, light, block and under factor, have very big difference, Image Feature Matching detects matching The number of point directly affects the accuracy of matching result, carries out by existing algorithm for the characteristic point of traditional implicit assumption is met Match somebody with somebody, for the region for being unsatisfactory for assumed condition, that is, the region for being difficult to direct matching is matched by the method for optimizing, obtained More match points, using SIFT algorithms to feature point detection and matching are carried out between adjacent two field picture, then using driver Colour of skin priori, is filtered and is tracked to characteristic matching result, obtains the human face region between different images, experimental result Show, on the image of 70 × 80 pixels, be obtained in that 150 characteristic points；

(4) head pose estimation：Based on the characteristic matching result in human face region, two width figures are estimated using the method for Epipolar geometry Attitude angle as between, i.e., a certain characteristic point X in three dimensions are projected in two different visual angles, imaging point position x1, X2, and imaging point is only relevant with the relative attitude between the parameter of video camera and camera, this paper driver head's attitudes, that is, consider head Relative motion between portion and photographic head, then meet：x^T ₂Fx₁=0

F is fundamental matrix, is the Algebraic Expression of limit restraint；E1 is set, and e2 is antipodal points, i.e. the baseline of two video cameras and imaging The intersection point of plane；C1, C2 are the center of two cameras；1 ' represents the x1 points in image, corresponding to pole in other image Line, i.e.,

1 '=FX₁

Basis matrix is obtained according to normalized linear 8 points of algorithms first, driver head's attitude angle is carried out on this basis, If the intrinsic parameter of video camera is K, then camera matrix is：

P=K [P/t]

X=PX is a bit on image, then there is K^-1So that,

$x^=K^{1}x$

Then pictures of 1 point of the space X under video camera normalization matrix is, basis matrix corresponding with normalization camera matrix is Video camera essential matrix with the relation of fundamental matrix is：

E=K^TFK

Essential matrix contains rotation and translates Vector Message, i.e.,：

E=[t]_xR

Spin matrix R can be obtained from essential matrix by above formula, spin matrix can be converted into Eulerian angles, i.e., rotating around Z, Y, X The anglec of rotation be α, beta, gamma, spin matrix can be expressed as：

$R=\left[\begin{array}{ccc}\cos \mathrm{\α}& sin\mathrm{\α}& 0\\ sin\mathrm{\α}& \cos \mathrm{\α}& 0\\ 0& 0& 1\end{array}\right]\left[\begin{array}{ccc}\cos \mathrm{\β}& 0& sin\mathrm{\β}\\ 0& 1& 0\\ sin\mathrm{\β}& 0& \cos \mathrm{\β}\end{array}\right]\left[\begin{array}{ccc}1& 0& 0\\ 0& cos\mathrm{\γ}& sin\mathrm{\γ}\\ 0& sin\mathrm{\γ}& \cos \mathrm{\γ}\end{array}\right]$

Based on the characteristic matching result in human face region, using dynamic Attitude estimation algorithm estimate basis matrix between image and Essential matrix, and then estimate spaced winding rotary shaft X of any two width adjacent image frame, the relative angle in tri- directions of Y, Z；

(5) experimental configuration analysis：During for carrying out head pose estimation using the system, the base obtained after being got on the bus using driver Quasi- image is standard front face face coordinate system, if the center of driver's inner eye corner line is zero, by IP Camera The each two field picture for obtaining, obtains enough characteristic point by Image Feature Matching algorithm, using RANSAC algorithms by mistake Match point filter, then using Algorithm of Head Pose Estimation, calculate driver's face relative to conventional coordinates spin moment Battle array, and then the α of driver head's attitude relative to frontal pose of the frame is estimated, the beta, gamma anglec of rotation works as driver head Roll attitude, i.e., now driver eyes to the front the most of the time, and head has left-right rotation by a small margin, i.e. roll, sees Road conditions are examined, the driver of record is in fatigue state, in driving procedure, occur that head is significantly low at set intervals, Overlook, then can lift rapidly, and the frequency more and more higher significantly nodded, in driver head's long-time left avertence, i.e., 120 frames are crossed in driftage, persistent period length, represent that driver is absent-minded or sleepy, occur overlooking and going off course situation, and system can send police Report, reminds driver.