CN107330378A - A kind of driving behavior detecting system based on embedded image processing - Google Patents
A kind of driving behavior detecting system based on embedded image processing Download PDFInfo
- Publication number
- CN107330378A CN107330378A CN201710433858.0A CN201710433858A CN107330378A CN 107330378 A CN107330378 A CN 107330378A CN 201710433858 A CN201710433858 A CN 201710433858A CN 107330378 A CN107330378 A CN 107330378A
- Authority
- CN
- China
- Prior art keywords
- image
- detection
- face
- driving behavior
- unit
- 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
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
- 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
- G06V40/10—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
- G06V40/172—Classification, e.g. identification
Landscapes
- Engineering & Computer Science (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)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of driving behavior detecting system based on embedded image processing, including：Main control module, information transmission modular, functional module, alarm module；Information transmission modular includes image/video transmission unit and driving behavior status signal transmission unit；Functional module drives detection unit and in violation of rules and regulations driving detection unit including image pre-processing unit, image/video memory cell, fatigue driving detection unit, blind regard；Functional module carries out image, video processing and analysis, and produces driving behavior status signal, and alarm module analysis driving behavior status signal carries out criticality assessment, decides whether to start alarm and selects alert levels according to analysis result.Beneficial effect：The a variety of behaviors of driver and state can be detected, detection range is wide, and strong applicability can effectively remind driver to note bad steering behavior and driving condition, the generation effectively cutd down traffic accidents.
Description
Technical field
The present invention relates to technical field of image processing, more particularly, to a kind of driver's row based on embedded image processing
For detecting system.
Background technology
Fatigue detecting system in current highway communication turns into domestic and international study hotspot, and fatigue driving also exists in length
With the train driver of way.With the quick increase of vehicle guaranteeding organic quantity and developing rapidly for highway, traffic accident is frequent
Occur, huge property loss and casualties is caused to countries in the world.How to reduce traffic accident have become it is worldwide
Problem.The reason for getting into an accident all is often driving lack of standardization and fatigue driving because driver, and existing detection is not advised
Model drives and the technology of fatigue driving often detects that accuracy is inadequate, detection range is limited, unsafe driving factor is considered not
Comprehensively.
The content of the invention
It is an object of the invention to overcome above-mentioned technical deficiency, a kind of driver's row based on embedded image processing is proposed
For detecting system, above-mentioned technical problem of the prior art is solved.
To reach above-mentioned technical purpose, technical scheme provides a kind of driver based on embedded image processing
Behavioral value system, including：Main control module, information transmission modular, functional module, alarm module；Information transmission modular includes figure
As Video Transmission Unit and driving behavior status signal transmission unit；Functional module includes image pre-processing unit, image/video
Memory cell, fatigue driving detection unit, blind regard drive detection unit and in violation of rules and regulations driving detection unit；Main control module connects each
Module and unit simultaneously coordinate the communication between modules and unit, and image/video transmission unit reads what infrared camera was shot
Image and video are simultaneously transferred to functional module, and functional module carries out image, video processing and analysis, and produces driving behavior state
Signal, the driving behavior status signal of driving behavior status signal transmission unit receiving module generation is simultaneously transferred to alarm mould
Block, alarm module analysis driving behavior status signal carries out criticality assessment, decides whether to start alarm and according to analysis
As a result alert levels are selected, and upload relevant image, video to server.
Compared with prior art, beneficial effects of the present invention include：Blink detection can be carried out, eye eye closing detection is narrowed, beats
Yawn detection, sight deviation detection, turn one's head bow detection, body body of one who has dropped dead by the roadside transfer detection, steering wheel slip out of the hand detection, safety belt examine
Survey, make a phone call to detect, you can a variety of behaviors of driver and state are detected, detection range is wide, under different illumination conditions
It can detect, strong applicability, can effectively remind driver to note bad steering behavior and driving condition, effectively cut down traffic accidents
Generation.
Brief description of the drawings
Fig. 1 is the external hardware framework for the driving behavior detecting system based on embedded image processing that the present invention is provided
Figure；
Fig. 2 is the system construction drawing for the driving behavior detecting system based on embedded image processing that the present invention is provided；
Fig. 3 is basic pretreatment operation flow chart；
Fig. 4 is KCF Face tracking algorithm flow charts；
Fig. 5 is active shape model ASM algorithm flow charts；
Fig. 6 is the shape regression algorithm overall flow figure of positioning feature point；
Fig. 7 is detection algorithm flow chart of bowing of turning one's head；
Fig. 8 is body body of one who has dropped dead by the roadside transfer detection algorithm flow chart；
Fig. 9 is detection algorithm flow chart of making a phone call.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
The present invention provides a kind of driving behavior detecting system based on embedded image processing, its external hardware facility
For：Infrared camera, embeded processor, user key-press, warning device and memory, wherein, infrared camera, user press
Key, warning device and memory with embeded processor connection, infrared camera obtain related face, steering wheel image or
Video information, by image or video information transmission to embeded processor, embeded processor is by corresponding algorithm to figure
As or video handled and analyzed, draw the current behavior state of driver, the violation of driver is driven, it is blind regard driving,
Fatigue driving state provides corresponding caution signal by warning device and reminds driver, and warning device is in driver without in violation of rules and regulations
Drive, blind understand stop alarm when regarding the state such as drivings, fatigue driving.
The present invention provides a kind of driving behavior detecting system based on embedded image processing, including：Main control module, letter
Cease transport module, functional module, alarm module；Information transmission modular includes image/video transmission unit and driving behavior state is believed
Number transmission unit；Functional module includes image pre-processing unit, image/video memory cell, fatigue driving detection unit, blind regarded
Drive detection unit and drive detection unit in violation of rules and regulations；Main control module connects modules and unit and coordinates modules and unit
Between communication, image/video transmission unit read infrared camera shoot image and video simultaneously be transferred to functional module, work(
Energy module carries out image, video processing and analysis, and produces driving behavior status signal, driving behavior status signal transmission unit
The driving behavior status signal of receiving module generation is simultaneously transferred to alarm module, alarm module analysis driving behavior state letter
Number criticality assessment is carried out, decide whether to start alarm and alert levels are selected according to analysis result, and upload relevant figure
Picture, video to server.
Driving behavior detecting system of the present invention based on embedded image processing, image pre-processing unit is carried out
Image/video collection, basic pretreatment, Face datection, face tracking, the pretreatment operation of facial modeling；Basic pre- place
The operation of reason includes：Change of scale, image gray processing, image filtering, image enhaucament, image binaryzation, Image Edge-Detection；People
The step of face is detected be：Using V-J method for detecting human face, face cycle detection is carried out to the image of input functional module, obtained
The rectangle frame of all doubtful face positions；The step of face tracking is：Using KCF track algorithms, input function is being obtained
During the Face datection result of the first two field picture of module, grader is trained, follow-up each two field picture is rolled up with the grader
Product, the maximum response of generation is face tracking result；The step of facial modeling is：Utilize active shape model
ASM algorithms or shape regression algorithm carry out facial modeling.
Substantially the operation pre-processed is specially：Each two field picture that image/video acquisition module is exported is carried out various pre-
Processing so that follow-up function algorithm processing is more in real time, accurately.Change of scale can reduce the resolution ratio of picture, accelerate follow-up
The speed of algorithm process；Image enhaucament can adjust gradation of image distribution, and reaching improves the purpose of picture contrast；Greyscale transformation
Triple channel image is transformed to single channel, accelerates subsequent algorithm processing speed；Image filtering carries out noise reduction process to image, reduces
Noise information is disturbed Processing Algorithm；Image binaryzation obtains the fore/background information of image, strengthens the description to target area；
Rim detection obtains the profile information of image object, strengthens the description to target shape.
The step of Face datection is specially：Using V-J method for detecting human face, face circulation inspection is carried out over an input image
Survey, and return to the rectangle frame of all doubtful face positions.Human face region is provided for follow-up facial modeling algorithm；
Initial tracing area is provided for Face tracking algorithm and it is periodically corrected；Positioning for driver's torso area provides people
Face positions reference information.
V-J arthmetic statements：V-J method for detecting human face has used Haar-like small on the basis of AdaBoost graders
Wave characteristic and integrogram method.AdaBoost is cascade classifier, and every grade is all retained into next with the discrimination being roughly the same
The candidate frame with face characteristic of level, and the sub-classifier per one-level is then made up of (by integral image meter many Haar features
Obtain, and save location), have horizontal, vertical, inclined, and each one threshold value of characteristic strip and two branch values,
And one total threshold value of every grade of sub-classifier band.When detecting face, integral image is equally calculated, is subsequent calculations Haar special
Levy and prepare, have the window an equal amount of window traversal entire image of face when then using with training, gradually later
Amplify window, equally do traversal search face；Whenever window is moved to a position, that is, the Haar features in the window are calculated,
Compared with the threshold value of Haar features in grader to select left or right branch value after weighting, the branch value for the level that adds up with
The threshold value of corresponding stage compares, and can just be screened more than the threshold value by entering next round.When by all grades of grader
Illustrate that the candidate frame there will be likely some face.
The step of face tracking is specially：Face tracking is the supplement of Face datection, can effectively eliminate Face datection
Shake, and improve the speed of Face datection.Face tracking is on the basis of Initial Face testing result is obtained, to follow-up
Image carries out human face region tracking.Human face region is provided for follow-up facial modeling algorithm and offer face (head) is complete
Whole movement locus.
The arthmetic statement of face tracking：Using KCF track algorithms.KCF is a kind of track algorithm rapidly and efficiently, and it is being carried
Take on the basis of face HOG features, realized and tracked using the principle of correlation filter.Algorithm extracts the HOG features of face first,
Grader is trained in the first frame of tracking, follow-up each two field picture is subjected to convolution, obtained maximum response with this grader
As tracking result, finally carries out grader renewal.Meanwhile, in order to which the drift for solving to produce during face tracking is asked
Topic, by the way of face tracking is combined, when tracking effect is less than threshold value, will carry out face and body shape using Face datection
The judgement of state, while every five frame carries out the repositioning of face.
Facial modeling：On the basis of Face datection, by the model trained in advance, face is further determined that
The position of characteristic point (eyes, eyebrow, nose, face, face's outline), according to the detection case of characteristic point, is determined whether
The reliability of Face datection result；By eyes and the feature point coordinates of face, the coarse positioning information of eyes and face can obtain；
According to the relative position relation between facial feature points, facial direction of visual lines can be estimated.
The arthmetic statement of facial modeling：
1st, active shape model ASM algorithms：
ASM algorithms are a kind of image search methods based on statistical model, by certain representational same class
Target object image is counted, so that obtain reacting the shape Statistics model of target object image two-dimensional shapes changing rule,
And the local texture model of the grey scale change rule of reflection characteristic point regional area.Known in target search procedure using priori
Know and carry out model initial alignment, then carry out feature point search using local texture model, and shape is entered using shape
Row fitting.ASM algorithms are divided into two steps of training and search.During training, the position constraint of each characteristic point is set up, each feature is built
The local feature of point.During search, Iterative matching.The first step：Training.Shape is built first；Then, it is each characteristic point structure
Build local feature.Purpose is that each characteristic point can find new position in each iterative search procedures.Local feature is general
With Gradient Features, to prevent illumination variation.Normal direction of some methods along edge is extracted, and some methods are near characteristic point
Extract rectangular area.Second step：Search.The position of eyes (or eyes and face) is calculated first, does simple yardstick and rotation
Transformationization, align face；Then each point after alignment is nearby searched for, match each local feature region (frequently with geneva away from
From), obtain preliminary configuration；Again with average face (shape) amendment matching result；Iteration is until convergence.In addition, frequently with
Multiple dimensioned method accelerates.The process of search is finally converged on high-resolution original image.
2nd, shape regression algorithm：
Position constraint between the textural characteristics of face and each characteristic point is combined.Using based on LBF textural characteristics knots
Machine learning algorithm-random forests algorithm is closed, design shape regression algorithm is positioned to human face characteristic point.Human face characteristic point is determined
Position is divided into two stages：1) off-line learning is training pattern；2) On-line matching is test model.
Shape regression algorithm is comprised the following steps that：
S21, preparation training sample：Sample includes Face Sample Storehouse and corresponding Face Sample Storehouse mark file ground
Truth；
S22, pretreatment training sample：Pretreatment operation is carried out to sample image, including (gray processing, image enhaucament,
Face datection), image cropping (quickening training speed)；
S23, to sample carry out coordinate transform, build average shape model：It is various in view of face between samples pictures
Various kinds, influenceed by each side such as illumination, postures, therefore when obtaining average shape, it should it is relatively uniform at one
Asked under framework, training sample is carried out after coordinate transform, average shape model is obtained to all characteristic point averageds；
S24, set training pattern parameter, including extract local binary feature random forest parameter (random tree number, with
Machine tree depth etc.) and global linear regression relevant parameter (return stage number of times, shape residual error etc.) for shape；
S25, training random forest, obtain Feature Mapping function, extract local binary feature, shape are carried out global linear
Return, and shape is updated, be specially：
First, random forest is trained, Feature Mapping function phi is obtainedt(I,St-1), extract local binary feature；
Then Δ S is usedt=WtΦt(I,St-1) global linear regression is carried out to shape, shape is updated, wherein,
ΔSt=Sgt-StFor the residual error in t-th of stage, W is linear regression matrix, and I is sample image；
S26, judge return number of times whether reach preset times, if being not reaching to preset times, circulation perform step
S25, if reaching preset times, preservation model.
Driving behavior detecting system of the present invention based on embedded image processing, fatigue driving detection unit is entered
Row blink detection operates, narrows eye eye closing detection operation；The step of blink detection is operated be：Image pre-processing unit is carried out to image
Pretreatment operation, extract face tracking obtain ocular, the ocular is the area-of-interest for including eyes, to this
Area-of-interest is pre-processed, and carries out average gray threshold binarization to pretreated area-of-interest, and fit eyes
Exact position, judge whether frequency of wink of blinking and calculate using upright projection；Narrowing the step of eye eye closing is detected is：Blinked
Eye detection operation, and the time persistently blinked is counted, it is judged as closing one's eyes if given threshold is exceeded.
Driving behavior detecting system of the present invention based on embedded image processing, fatigue driving detection unit is also
Progress, which is yawned, detects operation, and the step of detecting of yawning is：The pretreatment operation of image pre-processing unit is carried out to image, it is fixed
Position human face characteristic point, obtains mouth position and image, and edge analysis is carried out to mouth image, calculates the opening width of mouth and continues
Open the time judges whether driver yawns with this.
The step of blink detection is specially：The eye approximate region traced into is extracted first, and the region includes eyes
One big area-of-interest, makees the pretreatment such as appropriate cutting and size amplification to the area-of-interest.To pretreated
Eye areas carries out average gray threshold binarization, and goes out eyes using findContours and boundingRect Function Fittings
Exact position, judged whether in the exact position using upright projection blink.
The arthmetic statement of blink detection：The eyes approximate region traced into is extracted first, and the region is include eyes one
Individual big area-of-interest, in order to avoid the even interference brought to subsequent algorithm of uneven illumination, first can enter to area-of-interest
The appropriate cutting of row.Because the influence of illumination is easily brought on the left side of left eye and the right side of right eye, so right and left eyes are when cutting
Treat with a certain discrimination.Again due to some people eyebrow and eyes every it is close, may form UNICOM domain after binaryzation, influence eyes
Segmentation, so to carry out size amplification to the image cut, UNICOM domain can be eliminated substantially after twice of amplification, accurately by eye
Eyeball and eyebrow are distinguished.Two-value is carried out using average gray threshold value to last left eye and eye image after the completion of pre-processing above
Change, eyebrow and eyes become white after binaryzation, and remainder is black, then using findContours function lookups white
The profile in region, and by geometrical relationship, profile above is eyebrow, profile below is eyes.Find after profile, utilize
BoundingRect functions can go out the profile with rectangle fitting, extract the exact position that the rectangle is exactly eyes.In eyes
Judge whether blink in exact position using upright projection, what upright projection was calculated is that each row are black after eye areas binaryzation
The number of color pixel.Black picture element is distributed with very big difference in eye areas during due to eye opening and eye closing, and doing upright projection can
Intuitively to reflect very much, then by appropriate judgement sentence just can accurately detect blink.
Yawn detection the step of be specially：It is not accurate enough in order to avoid there is positioning feature point during positioning feature point
Situation, on the basis of positioning feature point, to mouth region respectively upwards downwards extension 15 pixels distance, in this, as
Mouth coarse positioning region, in order to further determine that the region of mouth, and can realize the analysis of mouth motion conditions, it is necessary to it
Fine positioning is carried out, in the case of daytime, carry out mouth fine positioning using half-tone information is easily influenceed by illumination variation, and mouth
Lip color is again comparatively larger with colour of skin difference, therefore carries out face fine positioning inspection using the lip color based on YIQ colour systems
Survey, and then state of yawning is judged by face splayed condition；For night, due to being that IMAQ is carried out under infrared environmental
Therefore in the absence of the influence of illumination variation, when mouth opens, mouth interior intensity value differs greatly compared with outside, therefore
Some pretreatments (including enhancing, filtering etc.) and adaptive threshold fuzziness are carried out to coarse positioning region, then chooses suitable strategy
Carry out contour detecting, it is possible to obtain the accurate location and shape of mouth, and then judge whether to yawn.
Yawn the arthmetic statement of detection：In 77 characteristic points obtained by facial modeling algorithm, 59-76 tables
Show the outline of face.We determine the rectangle in face region with four points therein (the left and right corners of the mouth and upper lower lip center)
Frame, and upward 15 pixels of extension downwards respectively on the region base.In the case of daytime, using YIQ colour systems
Carry out in lip color Threshold segmentation, YIQ colour systems, the optimal threshold of lip color is respectively Y ∈ [80,220], I ∈ [12,78], Q ∈
Lip and cavity interior are white in [7,25], the binary image after processing, and remainder is black；In the situation at night
Under, when mouth opens, mouth interior intensity value differs greatly compared with outside, therefore it is pre- to carry out some to coarse positioning region
Processing (including enhancing, filtering etc.) and max-thresholds segmentation, least square method then is carried out to the mouth area after segmentation oval
Fitting.Most can accurate description mouth motion conditions be exactly mouth the ratio of width to height change, people's mouth silent when is wide
Varying less for high ratio, is kept essentially constant；And when speaking or yawning with people, the wide height of current frame image mouth
It is constantly to change than the ratio of width to height with previous frame image.After by above-mentioned processing, it is possible to detection in real time
Mouth motion conditions, so that the change of mouth the ratio of width to height just can be accurately obtained very much, and then the motion conditions of determination mouth,
Such as speak or yawn.
Driving behavior detecting system of the present invention based on embedded image processing, it is blind to enter depending on driving detection unit
Row sight deviation detection, detection operation of bowing of turning one's head；The step of sight deviation detection is：Image pre-processing unit is carried out to image
Pretreatment operation, obtain all eye contour feature dot position informations and specific ear's characteristic point position information, definition
Line of sight parameters is ratio of the eye contour feature point barycenter to ear's characteristic point distance, according to line of sight parameters value and sight in image
The initialization value of parameter judges whether sight deviates；Turn one's head bow detection the step of be：Based on Face datection and face tracking, knot
Resultant motion detection positioning head region, the sight migration result obtained in conjunction with sight deviation detection judges whether driver has torsion
Head is bowed action.
Driving behavior detecting system of the present invention based on embedded image processing, blind regard drives detection unit also
Carry out body body of one who has dropped dead by the roadside transfer and detect that the step of operation, body body of one who has dropped dead by the roadside transfer detection is：Image pre-processing unit is carried out to image
Pretreatment operation, detects face and body region, and face and body region are carried out into ellipse fitting as overall, long by ellipse
Short axle, home position, deflection angle information judge body body of one who has dropped dead by the roadside transfer case.
The step of sight deviation detection is specially：Preceding 300 frame is initialized on the basis of it can detect face, for obtaining
Characteristic point near left eye portion at the barycenter of all characteristic points to left ear apart from l1With all characteristic points near right eye portion
Characteristic point at barycenter to auris dextra piece apart from l2Average of relatives value；After 300 frames, now above-mentioned face is calculated every time special
Distance ratio a little is levied, the ratio is compared with initializing obtained average value, if ratio is considered to right avertence more than 2.0
Turn；It is considered to deflect to the left if ratio is less than 0.5.
The arthmetic statement of sight deviation detection：Tested and found by a large amount of left and right head movements, when moving right, left eye portion
The barycenter apart from l1 than all characteristic points near right eye portion of characteristic point at the barycenter to left ear of neighbouring all characteristic points
Characteristic point to auris dextra piece it is much bigger apart from l2, therefore just using this, the ratio l1/l2 of both distances is used as the facial court in left and right
To the foundation of judgement.When experiment test finds 15 degree or so of deflection, both ratio is 2.0 or so and 0.5 or so.Pass through
The ratio is detected in real time, so as to judge sight direction by the size of ratio.
Turn one's head bow detection the step of be specially：Based on Face datection and face tracking, with reference to motion detection positioning head
Region.Judge whether driver turns one's head action of bowing in conjunction with the face orientation information in sight deviation detection module.Positioning
Face (head) region；Record the movement locus on head；With reference to the face orientation in head movement and sight deviation detection module
Information, comprehensive descision is turned one's head action of bowing.
Body body of one who has dropped dead by the roadside shifts the step of detecting：Human body region is extracted from background image from image
Come.On the basis of background modeling, image is converted into binary map, more complete is partitioned into sport foreground.Then, according to
It is identified go out human face region, acquiescence face below sport foreground be body region.It regard face and body region as entirety
Carry out ellipse fitting.The binary map of sport foreground is obtained by background modeling；It is true based on Face datection and human body topological structure
Determine face and body region；Face and body region are subjected to ellipse fitting as overall, by ellipse long and short shaft, home position,
The information such as deflection angle judge body body of one who has dropped dead by the roadside transfer case.
Driving behavior detecting system of the present invention based on embedded image processing, drives detection unit and enters in violation of rules and regulations
Line direction disk is slipped out of the hand detection operation, and steering wheel detecting step of slipping out of the hand is：The pretreatment behaviour that image pre-processing unit is carried out to image
Make, by hough ellipses detection outgoing direction disks position, accurately detect what is divided on steering wheel with reference to half-tone information and Skin Color Information
Single area-of-interest whether there is hand, so as to judge to whether there is manipulator on omnirange disk.
The slip out of the hand algorithm of detection of steering wheel is specially：
S1, one frame direction disk image of collection, detect that steering wheel image obtains steering wheel institute using hough Ellipses Detections
Ellipse, and determine 360 points on ellipse to be oval in the way of determining a point on ellipse at interval of 1 °
Line test point；
S2, the boundary rectangle from steering wheel image where interception steering wheel are used as the first picture, the whole of the first picture
Scope is area-of-interest, and area-of-interest is pre-processed；
S3, one single area-of-interest of acquisition, single area-of-interest are that ellipse inspection is first chosen on the first picture
Any one point in measuring point, then centered on the point of selection, interception size is size, and angle is used as single sense for 0 rectangle
Interest region, size size changes according to area-of-interest size adaptation；
S4, the binary image for obtaining the single area-of-interest obtained in S3, the binary image is represented in two dimension
Projected in coordinate system and to X-direction, the gray value for obtaining each row of the binary image is 255 pixel number A, and finding should
The specific continuous several columns of binary image and the length L for calculating continuous several columns, the binary image are specifically continuous
Several columns meet wherein each column pixel number A be all higher than first threshold T1, T1 according to size adaptive changes；
S5, the single area-of-interest progress Face Detection obtained to S3, obtain the binaryzation of the single area-of-interest
Hand skin color gray value is 255 in image, binary image, and identification ellipse test point falls the point in the single area-of-interest
Collection, the gray value for the point that test point is concentrated is 255 number N；
S6, work as L>Second Threshold T2 and N>During the 3rd threshold value T3, judge that the single area-of-interest that S3 is obtained has operation
Hand, otherwise, judges that manipulator is not present in the single area-of-interest that S3 is obtained, T2, T3 are according to size adaptive changes；
S7, circulation perform S3-S6 steps, and the central point using the S3 single area-of-interests obtained is with 1 ° as starting point
Step-length according to clockwise direction obtain single area-of-interest one by one, judge 360 centered on ellipse test point it is single
Area-of-interest whether there is manipulator；
Whether the single area-of-interest that S8, judgement have manipulator is continuous, and calculating continuously has the single of manipulator
The number B of area-of-interest, works as B>During the 4th threshold value T4, judge there is manipulator in the frame direction disk image gathered in S1,
Otherwise, judge manipulator is not present in the frame direction disk image gathered in S1, the central point of single area-of-interest is oval
Line test point, if the central point of some single area-of-interests is the continuous point on ellipse, judges that each single sense is emerging
Interesting region is continuous；
S9, circulation perform step S1-S8, the steering wheel image of collection setting frame number, the steering wheel figure of synthetic setting frame number
Manipulator's judged result of picture decides whether to remind driver.
Driving behavior detecting system of the present invention based on embedded image processing, drives detection unit also in violation of rules and regulations
Safety belt detection operation is carried out, safety belt detecting step is：The pretreatment operation of image pre-processing unit is carried out to image, to defeated
The image for entering functional unit carries out hough Line segment detections, and the line segment detected is screened, and detects whether to exist eligible
Line segment so as to judging whether driver with belt.
Safety belt detecting step and arthmetic statement：It whether there is safety belt straight line in detection image.Hough is carried out to image
Straight-line detection；The image detected is screened, detects whether there is safety belt straight line.
(1) hough detection of straight lines：Input as pretreated image, be output as lineSegments.
Hough straight-line detections are carried out to image, all straight line information detected are stored in container lineSegments,
The coordinate of two end points of the line segment detected is stored in lineSegments.
(2) safety belt is detected：Input as lineSegments, be output as identifying flag_safetybelt.
The line segment detected is screened, the length and angle of line segment are calculated by two extreme coordinates of line segment, and
And body rectangle frame this parameter for according to body couching that interface obtains, the line segment in body inframe is filtered out, further according to safety
The angle character and length characteristic of band, detect whether there is qualified safety belt straight line, if in the presence of, then it is assumed that driver system
There are safety belt, flag_safetybelt=1；If being not present, then it is assumed that driver does not fasten the safety belt, flag_safetybelt
=0.
Driving behavior detecting system of the present invention based on embedded image processing, drives detection unit also in violation of rules and regulations
Progress makes a phone call to detect operation, and the step of making a phone call and detect is：The pretreatment operation of image pre-processing unit is carried out to image, is led to
The binary map that background modeling obtains sport foreground is crossed, detection zone of making a phone call is positioned, detection lift hand, action of letting go, detection hand exist
Make a phone call region duration and judge whether driver is making a phone call.
Make a phone call detection the step of be specially：(1) binary map of sport foreground is obtained by background modeling；(2) positioning is beaten
Phone detection zone；(3) detection lift hand, action of letting go；(4) detection hand is in the duration for making a phone call region；(5) judge whether
Making a phone call.I.e.：Face datection is carried out first by Face datection and Face tracking algorithm and facial zone is positioned, according to fixed
The facial zone position and sizing that position goes out go out ear's image-region；It is inspection by the face both sides zone location adjacent with ear
Survey region.Face area is extended into detection zone, the feature of contained hand proportion in face extended area is extracted.With reference to
Steering wheel is slipped out of the hand the testing result of detection module, judges whether driver is making a phone call.
Compared with prior art, beneficial effects of the present invention include：Blink detection can be carried out, eye eye closing detection is narrowed, beats
Yawn detection, sight deviation detection, turn one's head bow detection, body body of one who has dropped dead by the roadside transfer detection, steering wheel slip out of the hand detection, safety belt examine
Survey, make a phone call to detect, you can a variety of behaviors of driver and state are detected, detection range is wide, under different illumination conditions
It can detect, strong applicability, can effectively remind driver to note bad steering behavior and driving condition, effectively cut down traffic accidents
Generation, and efficiency of algorithm, accuracy rate are high.
The embodiment of present invention described above, is not intended to limit the scope of the present invention..Any basis
Various other corresponding changes and deformation that the technical concept of the present invention is made, should be included in the guarantor of the claims in the present invention
In the range of shield.
Claims (10)
1. a kind of driving behavior detecting system based on embedded image processing, it is characterised in that including：Main control module, letter
Cease transport module, functional module, alarm module；Described information transport module includes image/video transmission unit and driving behavior shape
State signal transmission unit；The functional module includes image pre-processing unit, image/video memory cell, fatigue driving detection list
First, blind regard drives detection unit and in violation of rules and regulations driving detection unit；The main control module connection modules and unit simultaneously coordinate each
Communication between individual module and unit, described image Video Transmission Unit reads the image and video of infrared camera shooting and passed
Functional module is defeated by, the functional module carries out image, video processing and analysis, and produces driving behavior status signal, driven
The driving behavior status signal of behavior state signal transmission unit receiving module generation is simultaneously transferred to alarm module, the report
Alert module analysis driving behavior status signal carries out criticality assessment, decides whether to start alarm and is selected according to analysis result
Alert levels are selected, and upload relevant image, video to server.
2. the driving behavior detecting system as claimed in claim 1 based on embedded image processing, it is characterised in that described
Image pre-processing unit carries out image/video collection, basic pretreatment, Face datection, face tracking, facial modeling
Pretreatment operation；Substantially the operation pre-processed includes：Change of scale, image gray processing, image filtering, image enhaucament, image two
Value, Image Edge-Detection；The step of Face datection is：Using V-J method for detecting human face, the image of input functional module is entered
Pedestrian's face cycle detection, obtains the rectangle frame of all doubtful face positions；The step of face tracking is：Tracked using KCF
Algorithm, when obtaining the Face datection result of the first two field picture of input functional module, trains grader, will follow-up each frame figure
As carrying out convolution with the grader, the maximum response of generation is face tracking result；The step of facial modeling is：
Facial modeling is carried out using active shape model ASM algorithms or shape regression algorithm.
3. the driving behavior detecting system as claimed in claim 2 based on embedded image processing, it is characterised in that described
Fatigue driving detection unit carries out blink detection operation, narrows eye eye closing detection operation；The step of blink detection is operated be：To image
The pretreatment operation of image pre-processing unit is carried out, the ocular that face tracking is obtained is extracted, the ocular is to include eye
The area-of-interest of eyeball, is pre-processed to the area-of-interest, and average gray threshold value is carried out to pretreated area-of-interest
Binaryzation, and the exact position of eyes is fitted, judge whether frequency of wink of blinking and calculate using upright projection；Narrow eye eye closing
The step of detection is：Blink detection operation is carried out, and counts the time persistently blinked, is judged as closing if given threshold is exceeded
Eye.
4. the driving behavior detecting system as claimed in claim 2 based on embedded image processing, it is characterised in that described
Fatigue driving detection unit also carries out detection operation of yawning, and the step of detecting of yawning is：Image preprocessing is carried out to image
The pretreatment operation of unit, locating human face's characteristic point obtains mouth position and image, and edge analysis, meter are carried out to mouth image
Calculate the opening width of mouth and persistently open the time and judge whether driver yawns with this.
5. the driving behavior detecting system as claimed in claim 2 based on embedded image processing, it is characterised in that described
It is blind to carry out sight deviation detection depending on driving detection unit, turn one's head to bow and detect operation；The step of sight deviation detection is：To image
The pretreatment operation of image pre-processing unit is carried out, all eye contour feature dot position informations is obtained and specific ear is special
Dot position information is levied, it is ratio of the eye contour feature point barycenter to ear's characteristic point distance to define line of sight parameters, according to image
The initialization value of middle line of sight parameters value and line of sight parameters judges whether sight deviates；Turn one's head bow detection the step of be：Based on people
Face is detected and face tracking, with reference to motion detection positioning head region, is offset and is tied in conjunction with the sight that sight deviation detection is obtained
Fruit judges whether driver turns one's head action of bowing.
6. the driving behavior detecting system as claimed in claim 2 based on embedded image processing, it is characterised in that described
It is blind depending on drive detection unit also carry out body body of one who has dropped dead by the roadside transfer detection operation, body body of one who has dropped dead by the roadside transfer detection the step of be：Image is entered
The pretreatment operation of row image pre-processing unit, detects face and body region, and face and body region are entered as entirety
Row ellipse fitting, body body of one who has dropped dead by the roadside transfer case is judged by ellipse long and short shaft, home position, deflection angle information.
7. the driving behavior detecting system as claimed in claim 2 based on embedded image processing, it is characterised in that described
Drive detection unit travel direction disk in violation of rules and regulations to slip out of the hand detection operation, steering wheel detecting step of slipping out of the hand is：Image is carried out to image pre-
The pretreatment operation of processing unit, it is accurate with reference to half-tone information and Skin Color Information by hough ellipses detection outgoing direction disks position
The single area-of-interest divided on detection steering wheel whether there is hand, so as to judge to whether there is manipulator on omnirange disk.
8. the driving behavior detecting system as claimed in claim 2 based on embedded image processing, it is characterised in that described
Detection unit is driven in violation of rules and regulations and also carries out safety belt detection operation, and safety belt detecting step is：Image preprocessing list is carried out to image
The pretreatment operation of member, carries out hough Line segment detections to the image of input function unit, the line segment detected is screened,
Detect whether to have qualified line segment to judge whether driver is with belt.
9. the driving behavior detecting system as claimed in claim 2 based on embedded image processing, it is characterised in that described
Detection unit is driven in violation of rules and regulations and also carries out detection operation of making a phone call, and the step of making a phone call and detect is：Image preprocessing is carried out to image
The pretreatment operation of unit, the binary map of sport foreground is obtained by background modeling, positions detection zone of making a phone call, detection lift
Hand, action of letting go, detect hand in the duration in region of making a phone call and judge whether driver is making a phone call.
10. the driving behavior detecting system as claimed in claim 2 based on embedded image processing, it is characterised in that shape
Shape regression algorithm carry out facial modeling the step of be：Prepare training sample first, pre-process training sample, sample is entered
Row coordinate transform, builds average shape model, sets training pattern parameter, then carries out model training：Random forest is trained, is obtained
Feature Mapping function is obtained, local binary feature is extracted, global linear regression is carried out to shape, and shape is updated；Judge
Return whether number of times reaches preset times, if being not reaching to preset times, repeat model training, if reached default
Number of times, then preservation model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710433858.0A CN107330378A (en) | 2017-06-09 | 2017-06-09 | A kind of driving behavior detecting system based on embedded image processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710433858.0A CN107330378A (en) | 2017-06-09 | 2017-06-09 | A kind of driving behavior detecting system based on embedded image processing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107330378A true CN107330378A (en) | 2017-11-07 |
Family
ID=60195177
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710433858.0A Pending CN107330378A (en) | 2017-06-09 | 2017-06-09 | A kind of driving behavior detecting system based on embedded image processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107330378A (en) |
Cited By (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107862298A (en) * | 2017-11-27 | 2018-03-30 | 电子科技大学 | It is a kind of based on the biopsy method blinked under infrared eye |
| CN108309311A (en) * | 2018-03-27 | 2018-07-24 | 北京华纵科技有限公司 | A kind of real-time doze of train driver sleeps detection device and detection algorithm |
| CN108509902A (en) * | 2018-03-30 | 2018-09-07 | 湖北文理学院 | A kind of hand-held telephone relation behavioral value method during driver drives vehicle |
| CN108734086A (en) * | 2018-03-27 | 2018-11-02 | 西安科技大学 | The frequency of wink and gaze estimation method of network are generated based on ocular |
| CN108960065A (en) * | 2018-06-01 | 2018-12-07 | 浙江零跑科技有限公司 | A kind of driving behavior detection method of view-based access control model |
| CN109062561A (en) * | 2018-07-27 | 2018-12-21 | 淮海工学院 | One kind being based on Python dangerous driving early warning system |
| CN109214289A (en) * | 2018-08-02 | 2019-01-15 | 厦门瑞为信息技术有限公司 | A kind of Activity recognition method of making a phone call from entirety to local two stages |
| CN109360375A (en) * | 2018-11-26 | 2019-02-19 | 青岛小鸟看看科技有限公司 | A kind of method and system improving fatigue driving accuracy in detection |
| CN109472228A (en) * | 2018-10-29 | 2019-03-15 | 上海交通大学 | A kind of yawn detection method based on deep learning |
| CN109508089A (en) * | 2018-10-30 | 2019-03-22 | 上海大学 | A kind of sight control system and method based on level random forest |
| CN109508632A (en) * | 2018-09-30 | 2019-03-22 | 百度在线网络技术（北京）有限公司 | Image processing method, device, equipment and computer readable storage medium |
| CN110103820A (en) * | 2019-04-24 | 2019-08-09 | 深圳市轱辘汽车维修技术有限公司 | The method, apparatus and terminal device of the abnormal behaviour of personnel in a kind of detection vehicle |
| CN110293974A (en) * | 2019-07-02 | 2019-10-01 | 重庆大学 | Driving gesture recognition and management system and APP based on lightweight neural network are applied |
| WO2019188060A1 (en) * | 2018-03-30 | 2019-10-03 | 株式会社フジクラ | Detection device and detection method |
| CN110334600A (en) * | 2019-06-03 | 2019-10-15 | 武汉工程大学 | A kind of multiple features fusion driver exception expression recognition method |
| CN110386060A (en) * | 2018-04-20 | 2019-10-29 | 比亚迪股份有限公司 | Steering wheel both hands are detached from based reminding method, device, system and the vehicle with it |
| CN110550043A (en) * | 2019-09-05 | 2019-12-10 | 上海博泰悦臻网络技术服务有限公司 | Dangerous behavior warning method and system, computer storage medium and vehicle-mounted terminal |
| CN110909568A (en) * | 2018-09-17 | 2020-03-24 | 北京京东尚科信息技术有限公司 | Image detection method, apparatus, electronic device, and medium for face recognition |
| CN111532281A (en) * | 2020-05-08 | 2020-08-14 | 奇瑞汽车股份有限公司 | Driving behavior monitoring method and device, terminal and storage medium |
| CN111553190A (en) * | 2020-03-30 | 2020-08-18 | 浙江工业大学 | Image-based driver attention detection method |
| CN112092898A (en) * | 2019-06-17 | 2020-12-18 | 比亚迪股份有限公司 | Vehicle and control method and device thereof |
| CN112241647A (en) * | 2019-07-16 | 2021-01-19 | 青岛点之云智能科技有限公司 | Dangerous driving behavior early warning device and method based on depth camera |
| CN112307846A (en) * | 2019-08-01 | 2021-02-02 | 北京新联铁集团股份有限公司 | Analysis method for violation of crew service |
| CN112347820A (en) * | 2019-08-08 | 2021-02-09 | 株洲中车时代电气股份有限公司 | Driver driving behavior monitoring method and device |
| CN113126296A (en) * | 2020-01-15 | 2021-07-16 | 未来（北京）黑科技有限公司 | Head-up display equipment capable of improving light utilization rate |
| CN113239754A (en) * | 2021-04-23 | 2021-08-10 | 泰山学院 | Dangerous driving behavior detection and positioning method and system applied to Internet of vehicles |
| CN113343926A (en) * | 2021-07-01 | 2021-09-03 | 南京信息工程大学 | Driver fatigue detection method based on convolutional neural network |
| CN113619386A (en) * | 2021-08-19 | 2021-11-09 | 四川轻化工大学 | Embedded multidimensional perception driver safety assistance and alarm system |
| CN114743184A (en) * | 2022-06-10 | 2022-07-12 | 航天科技控股集团股份有限公司 | Driver driving state early warning system |
| CN118658148A (en) * | 2024-08-15 | 2024-09-17 | 名商科技有限公司 | User driving behavior detection method, system and storage medium |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101599207A (en) * | 2009-05-06 | 2009-12-09 | 深圳市汉华安道科技有限责任公司 | A kind of fatigue driving detection device and automobile |
| CN102263937A (en) * | 2011-07-26 | 2011-11-30 | 华南理工大学 | Driver's driving behavior monitoring device and monitoring method based on video detection |
| CN103150556A (en) * | 2013-02-20 | 2013-06-12 | 西安理工大学 | Safety belt automatic detection method for monitoring road traffic |
| JP2013123180A (en) * | 2011-12-12 | 2013-06-20 | Denso Corp | Monitoring device |
| CN103279750A (en) * | 2013-06-14 | 2013-09-04 | 清华大学 | Detecting method of mobile telephone holding behavior of driver based on skin color range |
| CN103413114A (en) * | 2013-05-17 | 2013-11-27 | 浙江大学 | Near-drowning behavior detection method based on support vector machine |
| CN105354987A (en) * | 2015-11-26 | 2016-02-24 | 南京工程学院 | Vehicle fatigue driving detection and identity authentication apparatus, and detection method thereof |
| CN106485214A (en) * | 2016-09-28 | 2017-03-08 | 天津工业大学 | A kind of eyes based on convolutional neural networks and mouth state identification method |
| CN106709420A (en) * | 2016-11-21 | 2017-05-24 | 厦门瑞为信息技术有限公司 | Method for monitoring driving behaviors of driver of commercial vehicle |
-
2017
- 2017-06-09 CN CN201710433858.0A patent/CN107330378A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101599207A (en) * | 2009-05-06 | 2009-12-09 | 深圳市汉华安道科技有限责任公司 | A kind of fatigue driving detection device and automobile |
| CN102263937A (en) * | 2011-07-26 | 2011-11-30 | 华南理工大学 | Driver's driving behavior monitoring device and monitoring method based on video detection |
| JP2013123180A (en) * | 2011-12-12 | 2013-06-20 | Denso Corp | Monitoring device |
| CN103150556A (en) * | 2013-02-20 | 2013-06-12 | 西安理工大学 | Safety belt automatic detection method for monitoring road traffic |
| CN103413114A (en) * | 2013-05-17 | 2013-11-27 | 浙江大学 | Near-drowning behavior detection method based on support vector machine |
| CN103279750A (en) * | 2013-06-14 | 2013-09-04 | 清华大学 | Detecting method of mobile telephone holding behavior of driver based on skin color range |
| CN105354987A (en) * | 2015-11-26 | 2016-02-24 | 南京工程学院 | Vehicle fatigue driving detection and identity authentication apparatus, and detection method thereof |
| CN106485214A (en) * | 2016-09-28 | 2017-03-08 | 天津工业大学 | A kind of eyes based on convolutional neural networks and mouth state identification method |
| CN106709420A (en) * | 2016-11-21 | 2017-05-24 | 厦门瑞为信息技术有限公司 | Method for monitoring driving behaviors of driver of commercial vehicle |
Non-Patent Citations (3)
| Title |
|---|
| 周民刚: ""基于计算机视觉的人体跌倒检测算法研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 张恒: "基于近红外图像的疲劳驾驶检测研究与系统实现", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 王威等: "一种用于驾驶员疲劳检测的人眼状态判别方法", 《计算机光盘软件与应用》 * |
Cited By (38)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107862298A (en) * | 2017-11-27 | 2018-03-30 | 电子科技大学 | It is a kind of based on the biopsy method blinked under infrared eye |
| CN107862298B (en) * | 2017-11-27 | 2021-07-06 | 电子科技大学 | Winking living body detection method based on infrared camera device |
| CN108309311A (en) * | 2018-03-27 | 2018-07-24 | 北京华纵科技有限公司 | A kind of real-time doze of train driver sleeps detection device and detection algorithm |
| CN108734086A (en) * | 2018-03-27 | 2018-11-02 | 西安科技大学 | The frequency of wink and gaze estimation method of network are generated based on ocular |
| CN108734086B (en) * | 2018-03-27 | 2021-07-27 | 西安科技大学 | Blink frequency and sight line estimation method based on eye area generation network |
| CN108509902A (en) * | 2018-03-30 | 2018-09-07 | 湖北文理学院 | A kind of hand-held telephone relation behavioral value method during driver drives vehicle |
| CN108509902B (en) * | 2018-03-30 | 2020-07-03 | 湖北文理学院 | Method for detecting call behavior of handheld phone in driving process of driver |
| WO2019188060A1 (en) * | 2018-03-30 | 2019-10-03 | 株式会社フジクラ | Detection device and detection method |
| CN110386060A (en) * | 2018-04-20 | 2019-10-29 | 比亚迪股份有限公司 | Steering wheel both hands are detached from based reminding method, device, system and the vehicle with it |
| CN108960065A (en) * | 2018-06-01 | 2018-12-07 | 浙江零跑科技有限公司 | A kind of driving behavior detection method of view-based access control model |
| CN108960065B (en) * | 2018-06-01 | 2020-11-17 | 浙江零跑科技有限公司 | Driving behavior detection method based on vision |
| CN109062561A (en) * | 2018-07-27 | 2018-12-21 | 淮海工学院 | One kind being based on Python dangerous driving early warning system |
| CN109214289B (en) * | 2018-08-02 | 2021-10-22 | 厦门瑞为信息技术有限公司 | Method for recognizing two-stage calling behavior from whole to local |
| CN109214289A (en) * | 2018-08-02 | 2019-01-15 | 厦门瑞为信息技术有限公司 | A kind of Activity recognition method of making a phone call from entirety to local two stages |
| CN110909568A (en) * | 2018-09-17 | 2020-03-24 | 北京京东尚科信息技术有限公司 | Image detection method, apparatus, electronic device, and medium for face recognition |
| CN109508632A (en) * | 2018-09-30 | 2019-03-22 | 百度在线网络技术（北京）有限公司 | Image processing method, device, equipment and computer readable storage medium |
| CN109508632B (en) * | 2018-09-30 | 2023-04-07 | 百度在线网络技术（北京）有限公司 | Image processing method, device, equipment and computer readable storage medium |
| CN109472228A (en) * | 2018-10-29 | 2019-03-15 | 上海交通大学 | A kind of yawn detection method based on deep learning |
| CN109508089A (en) * | 2018-10-30 | 2019-03-22 | 上海大学 | A kind of sight control system and method based on level random forest |
| CN109360375A (en) * | 2018-11-26 | 2019-02-19 | 青岛小鸟看看科技有限公司 | A kind of method and system improving fatigue driving accuracy in detection |
| CN110103820A (en) * | 2019-04-24 | 2019-08-09 | 深圳市轱辘汽车维修技术有限公司 | The method, apparatus and terminal device of the abnormal behaviour of personnel in a kind of detection vehicle |
| CN110334600A (en) * | 2019-06-03 | 2019-10-15 | 武汉工程大学 | A kind of multiple features fusion driver exception expression recognition method |
| CN112092898A (en) * | 2019-06-17 | 2020-12-18 | 比亚迪股份有限公司 | Vehicle and control method and device thereof |
| CN110293974A (en) * | 2019-07-02 | 2019-10-01 | 重庆大学 | Driving gesture recognition and management system and APP based on lightweight neural network are applied |
| CN112241647A (en) * | 2019-07-16 | 2021-01-19 | 青岛点之云智能科技有限公司 | Dangerous driving behavior early warning device and method based on depth camera |
| CN112241647B (en) * | 2019-07-16 | 2023-05-09 | 青岛点之云智能科技有限公司 | Dangerous driving behavior early warning device and method based on depth camera |
| CN112307846A (en) * | 2019-08-01 | 2021-02-02 | 北京新联铁集团股份有限公司 | Analysis method for violation of crew service |
| CN112347820A (en) * | 2019-08-08 | 2021-02-09 | 株洲中车时代电气股份有限公司 | Driver driving behavior monitoring method and device |
| CN110550043A (en) * | 2019-09-05 | 2019-12-10 | 上海博泰悦臻网络技术服务有限公司 | Dangerous behavior warning method and system, computer storage medium and vehicle-mounted terminal |
| CN113126296B (en) * | 2020-01-15 | 2023-04-07 | 未来（北京）黑科技有限公司 | Head-up display equipment capable of improving light utilization rate |
| CN113126296A (en) * | 2020-01-15 | 2021-07-16 | 未来（北京）黑科技有限公司 | Head-up display equipment capable of improving light utilization rate |
| CN111553190A (en) * | 2020-03-30 | 2020-08-18 | 浙江工业大学 | Image-based driver attention detection method |
| CN111532281A (en) * | 2020-05-08 | 2020-08-14 | 奇瑞汽车股份有限公司 | Driving behavior monitoring method and device, terminal and storage medium |
| CN113239754A (en) * | 2021-04-23 | 2021-08-10 | 泰山学院 | Dangerous driving behavior detection and positioning method and system applied to Internet of vehicles |
| CN113343926A (en) * | 2021-07-01 | 2021-09-03 | 南京信息工程大学 | Driver fatigue detection method based on convolutional neural network |
| CN113619386A (en) * | 2021-08-19 | 2021-11-09 | 四川轻化工大学 | Embedded multidimensional perception driver safety assistance and alarm system |
| CN114743184A (en) * | 2022-06-10 | 2022-07-12 | 航天科技控股集团股份有限公司 | Driver driving state early warning system |
| CN118658148A (en) * | 2024-08-15 | 2024-09-17 | 名商科技有限公司 | User driving behavior detection method, system and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107330378A (en) | A kind of driving behavior detecting system based on embedded image processing | |
| CN112101175B (en) | Expressway vehicle detection and multi-attribute feature extraction method based on local image | |
| CN103400110B (en) | Abnormal face detecting method before ATM cash dispenser | |
| CN106203385B (en) | A kind of hand-held phone behavioral value method and device of driver | |
| CN102289660B (en) | Method for detecting illegal driving behavior based on hand gesture tracking | |
| CN105260699B (en) | A kind of processing method and processing device of lane line data | |
| CN107292251B (en) | Driver fatigue detection method and system based on human eye state | |
| CN108985230A (en) | Method for detecting lane lines, device and computer readable storage medium | |
| CN100452081C (en) | Human eye positioning and human eye state recognition method | |
| CN109670396A (en) | A kind of interior Falls Among Old People detection method | |
| CN106250801A (en) | Based on Face datection and the fatigue detection method of human eye state identification | |
| CN109635875A (en) | A kind of end-to-end network interface detection method based on deep learning | |
| CN104463877B (en) | A kind of water front method for registering based on radar image Yu electronic chart information | |
| CN111553214B (en) | Method and system for detecting smoking behavior of driver | |
| CN104091147A (en) | Near infrared eye positioning and eye state identification method | |
| CN106326858A (en) | Road traffic sign automatic identification and management system based on deep learning | |
| CN104715238A (en) | Pedestrian detection method based on multi-feature fusion | |
| CN110232389A (en) | A kind of stereoscopic vision air navigation aid based on green crop feature extraction invariance | |
| CN101715111B (en) | Method for automatically searching abandoned object in video monitoring | |
| CN106846734A (en) | A kind of fatigue driving detection device and method | |
| CN108985170A (en) | Transmission line of electricity hanger recognition methods based on Three image difference and deep learning | |
| CN109711264A (en) | A kind of bus zone road occupying detection method and device | |
| CN103310194A (en) | Method for detecting head and shoulders of pedestrian in video based on overhead pixel gradient direction | |
| CN112381870B (en) | Binocular vision-based ship identification and navigational speed measurement system and method | |
| CN106371148A (en) | Millimeter wave image-based human body foreign substance detection method and system |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171107 |