CN107226023A - Image procossing real-time early warning system and method - Google Patents
Image procossing real-time early warning system and method Download PDFInfo
- Publication number
- CN107226023A CN107226023A CN201710228535.8A CN201710228535A CN107226023A CN 107226023 A CN107226023 A CN 107226023A CN 201710228535 A CN201710228535 A CN 201710228535A CN 107226023 A CN107226023 A CN 107226023A
- Authority
- CN
- China
- Prior art keywords
- image
- module
- unit
- vehicle
- surrounding vehicles
- 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
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60Q—ARRANGEMENT OF SIGNALLING OR LIGHTING DEVICES, THE MOUNTING OR SUPPORTING THEREOF OR CIRCUITS THEREFOR, FOR VEHICLES IN GENERAL
- B60Q9/00—Arrangement or adaptation of signal devices not provided for in one of main groups B60Q1/00 - B60Q7/00, e.g. haptic signalling
- B60Q9/008—Arrangement or adaptation of signal devices not provided for in one of main groups B60Q1/00 - B60Q7/00, e.g. haptic signalling for anti-collision purposes
-
- 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
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Abstract
The present invention relates to a kind of image procossing real-time early warning system and method;Including with lower module, the sensing module for monitoring surrounding vehicles running status and shape;For the processing module calculated according to sensing module monitoring information;The alarm module warned for being obtained a result according to processing module;Power module for providing sensing module, processing module and alarm module supply of electric power;The surrounding vehicles monitored and the various information of vehicles stored in memory cell are compared to the performance for determining the surrounding vehicles, show whether the car is dangerous to body vehicle by computing.
Description
Technical field
The present invention relates to a kind of traffic safety early warning system, and in particular to a kind of image procossing real-time early warning system and side
Method.
Background technology
With being continuously increased for vehicle fleet size, traffic safety becomes more and more important in daily life, therefore automobile is endangered
Dangerous early warning system becomes extremely important, in the prior art, it is general using laser range finder carry out distance perspective should and early warning, but specialty
Level high-precision three-dimensional laser range finder is expensive, and is not suitable for common vehicle identification vehicle, ranging, early warning;And it is common
The applicable accuracy of laser distance measuring instrument of masses is not high, and disturbing factor is a lot, and data are unreliable.
The content of the invention
In order to solve the above problems, there is provided a kind of higher early warning of simple in construction, easy to use, cheap, accuracy
System, the present invention devises a kind of image procossing real-time early warning system and method.
The concrete technical scheme taken of the present invention is:A kind of image procossing real-time early warning system, including with lower module, use
In monitoring surrounding vehicles running status and the sensing module of shape;For the processing calculated according to sensing module monitoring information
Module;The alarm module warned for being obtained a result according to processing module;For to sensing module, processing module and report
The power module of alert modules with power supply;
Sensing module includes being provided with timing transmitting element in camera and laser range finder, sensing module;
Processing module includes, the image border processing unit for carrying out marginalisation processing to surrounding vehicles image;With
In the areal calculation unit for calculating unit interval inward flange image area accounting variation tendency;For with rate calculation unit
Synchronously draw the distance computation unit of body vehicle and surrounding vehicles relative distance variation tendency;For by areal calculation unit with
Distance computation unit show that data calculate the rate calculation unit of movement rate in the surrounding vehicles unit interval;For
RATES's unit that contrast draws relative motion state is carried out with body car speed;For according to rate calculation unit and speed
Rate comparison unit, which is obtained a result, carries out the dangerous budget unit of budget to danger coefficient;For storing vehicle and all kinds of vehicle information
Memory cell;The Model Matching list that picture after for handling marginalisation is matched with the progress that memory cell stores up information
Member.
Preferably, alarm module is to be provided with to send dangerous budget unit result to alarm mould in loudspeaker, processing module
The voice driven unit of block.
Preferably, the present apparatus is arranged on automobile, and some sensing modules are correspondingly arranged at motor vehicle environment.Laser range finder bag
Include Laser emission end, laser pick-off end.
Preferably, the image partition holding of the monitoring image for sensing module is provided with memory cell;In processing module
Unit is cleared up provided with the timing for image partition holding.
A kind of image procossing real time early warning method, comprises the following steps,
A. surrounding vehicles are monitored by sensing module, and the image and range information that sensing module is monitored
Timing transmission is to processing module;
B. image border processing unit carries out marginalisation processing, unit of account time inner edge to the image information received
The variation tendency of edge image relative image gross area accounting size, and body car is synchronously drawn according to sensing module Monitoring Data
Relative spacing variation tendency with surrounding vehicles;
C. set in the database of storage vehicle and all kinds of vehicle information, the figure and database that marginalisation image is drawn
Data carry out contrast and draw the affiliated type of surrounding vehicles;
D. variation tendency and spacing variation tendency are accounted for according to area and fortune in the surrounding vehicles unit interval is drawn by computing
Dynamic speed;
E. body state of motion of vehicle information is imported, with reference to the step d bodies drawn and in the surrounding vehicles unit interval
Relative movement information, carry out danger coefficient and move the result that may occur according to current state to enter according to relative motion relation
Row budget;
F. danger coefficient and the result that may occur are sent to alarm module.
Preferably, the image partition holding of the monitoring image for sensing module is provided with database;To image storage point
The picture of area's memory storage is timed cleaning.
Preferably, this method is used for vehicle early warning, alarm module is loudspeaker.
Further, the range information monitored and image information are transmitted by timing transmitting element timing, work as discovery
Surrounding vehicles and body vehicle distances it is small to a fixed value when, monitor and transmit in real time.
Further, the model of all angles for the basic vehicle that is stored with memory cell, including two wheels, three-wheel, four-wheel
Headstock, the tailstock, strabismus model etc.;In addition, each type of vehicle performance parameter of correspondence, such as power property arguments are included
(max. speed, braking time etc.), structural parameters (overall width, vehicle commander etc.).
Further, sensing module is arranged on the place that vehicle chamber interior energy sees surrounding clearly, or sensing module is located at vehicle
Outside surrounding, the car generally taken shelter from rain under, installation region be the blind spot of human eye, such as car after, the car left and right sides etc..
Advantageous effects:It is simple in construction, drawn by image change trend and the relative distance change to surrounding vehicles
The speed of the car, imports the motion state of body vehicle, by the relative motion state for calculating surrounding vehicles;It will monitor
Surrounding vehicles and memory cell in the various information of vehicles that are stored the performance for determining the surrounding vehicles is compared, pass through fortune
Calculation show whether the car is dangerous to body vehicle, plays out information warning by alarm module, effectively reminds driver
Note eliminating driving blind area vehicle, it is to avoid occur unexpected traffic accident.
Brief description of the drawings
Fig. 1 is the system structure diagram;
Fig. 2 is the system module sketch;
Fig. 3 is operating process sketch of the present invention.
Marginal data:
The body vehicle 1 of the system is set
Surrounding vehicles 2
Sensing module 3
Embodiment
A kind of image procossing real-time early warning system of embodiment one, referring to Fig. 1-3:Including with lower module, around monitoring
The sensing module of travel condition of vehicle and shape;For the processing module calculated according to sensing module monitoring information;For
Obtained a result according to processing module the alarm module warned;For being carried to sensing module, processing module and alarm module
For the power module of supply of electric power;
The sensing module includes being provided with timing transmitting element in camera and laser range finder, sensing module;
The processing module includes, and handles single for carrying out the image borderization of marginalisation processing to surrounding vehicles image
Member;Areal calculation unit for calculating unit interval inward flange image area accounting variation tendency;
By with rate calculation unit synchronously draw the spacing of body vehicle and surrounding vehicles relative distance variation tendency based on
Calculate unit;For areal calculation unit and distance computation unit to be drawn into data calculate in the surrounding vehicles unit interval
The rate calculation unit of movement rate;Show that the RATES of relative motion state is single for carrying out contrast with body car speed
Member;The dangerous budget list of budget is carried out to danger coefficient for being obtained a result according to rate calculation unit and RATES's unit
Member;Memory cell for storing vehicle and all kinds of vehicle information;
The Model Matching unit that picture after for handling marginalisation is matched with the progress that memory cell stores up information.
Further, the range information monitored and image information are transmitted by timing transmitting element timing, work as discovery
Surrounding vehicles and body vehicle distances it is small to a fixed value when, monitor and transmit in real time.
Further, the model of all angles for the basic vehicle that is stored with memory cell, including two wheels, three-wheel, four-wheel
Headstock, the tailstock, strabismus model etc.;In addition, each type of vehicle performance parameter of correspondence, such as power property arguments are included
(max. speed, braking time etc.), structural parameters (overall width, vehicle commander etc.).
Further, sensing module, which is disposed in the interior, can see the place of surrounding clearly, or sensing module is located at the outside of car
Surrounding, the car generally taken shelter from rain under, installation region be the blind spot of human eye, such as car after, the car left and right sides etc..
Image edge processing unit carries out rim detection and extraction to image, comprises the following steps:Gaussian function is used first
Several be filtered to image obtains a smoothed data array, carries out gradient calculation, can be with after the direction of gradient is aware of
By edge direction with 45 ° of divisions, eight directions, by gradient direction, the adjacent pixels in this pixel gradient direction can be found.
In order to ensure edge accurate positioning, and keep single-phase to answer, the ridge band in gradient magnitude image is refined, retain width
It is worth the maximum point of localized variation, during non-maximum restraining, for the ladder obtained in each point and previous step of image
Direction is spent, it is compared with the value on gradient direction, if it is lower, illustrate that the point is not local maximum, then will correspondence
Edge be expressed as 0, i.e., the point is not marginal point.Set Low threshold and high threshold that dual threshold method detection and connection edge need
Value, dual threashold value processing is carried out to gradient image, and what gradient magnitude was more than high threshold is edge, and gradient magnitude is less than Low threshold
It is not edge, what gradient magnitude fell between, judge to whether there is more than high threshold in eight field pixels of the pixel
Edge pixel, if in the presence of if it be edge pixel, be not otherwise to determine after a marginal point, continuous search and track edge, marginal point
After connecting;By the edge extracting of whole image, the image in image information is divided by the marginalisation of extraction, it is real
The segmentation of existing object and background.
The image that Model Matching unit handles marginalisation carries out the identification of type of vehicle, and specific steps include:To car
Image can obtain vehicle to external boundary after treatment, it is shown that the shape of vehicle, in a closed curve.One is closed
Curve took N number of sampled point by equal intervals, obtains coordinate sequence, and the coordinate sequence is periodic function, with Zhou Changwei weeks of curve
Phase.Fast discrete Fourier transformation is carried out to coordinate sequence in a cycle, the motion of vehicle and the visual field of camera can make
Into the translation of above-mentioned curve, amplification and the change of FFT starting points, it is therefore desirable to extract the ginseng that curve does not change with above-mentioned factor
Number.With reference to the data of memory cell, when dispersion level off to 0 when, represent that two cars shape is basically identical;Dispersion is bigger, says
The difference of bright vehicle shape is bigger.
Exist in the areal calculation unit unit of account time per the area x of the surrounding vehicles image of two field picture marginalisation processing
Shared area ratio in whole image area y, and then the unit interval interior area is counted than f (x, y), analyze the area and compare f
The size variation trend f ' (x, y) of (x, y).
Vehicle distance data g (z) in the distance computation unit statistical unit time, analyzes the size variation trend of the vehicle distance data
g’(z)。
Rate calculation unit is used for the relative speed for obtaining surrounding vehicles and body vehicle, compare the area than size become
The size variation of change trend f ' (x, y) and the vehicle distance data becomes g ' (z), if both differed within the unit interval 0~10% with
It is interior, then appoint and take a value, try to achieve the rate value V within the unit interval.If both are differed by more than within 10% within the unit interval, take
Both average value V=[f ' (x, y)+g ' (z)]/2, try to achieve the rate value V within the unit interval;
RATES's unit is used for the speed for obtaining surrounding vehicles, imports velocity amplitude f of this car within the unit interval
(v), if V values are more than 1, then it represents that other side is close, and speed is V1=f (v)+V values, and direction is arranged by specific sensing module
Depending on position.If V values are less than 1, then it represents that other side is remote, and speed is V2=f (v)-V values, and specific sensing module is pressed in direction
Depending on the position of arrangement.If V values are equal to 1, then it represents that B cars are identical with A vehicle speeds.
Dangerous budget unit binding model matching unit contrasts the performance parameter of successful vehicle and the car that monitors
Speed, the travel direction of the car, calculating obtain braking distance, by the braking distance normal spacing progress pair corresponding with the speed
Than, and calculate by travelling following how long interior may occur the data results such as accident at this rate.It would be possible to occur thing
Therefore result of calculation, combined with the motion state of this car progress voice broadcast early warning.Result of calculation includes:Type of vehicle and sheet
Car travel direction, speed, the estimated accident that may occur by the car motion state.
The present invention is simple in construction, and the speed of the car is drawn by image change trend and the relative distance change to surrounding vehicles
Rate value, introduces the speed of body vehicle, by the direction of relative movement for calculating surrounding vehicles;By the surrounding vehicles monitored
The performance for determining the surrounding vehicles is compared with the various information of vehicles that are stored in memory cell, the car is drawn by computing
It is whether dangerous to body vehicle, information warning is played out by alarm module, effectively reminds driver to note driving blind
Area's vehicle, it is to avoid occur unexpected traffic accident.
A kind of image procossing real-time early warning system of embodiment two, referring to Fig. 1-3:Including with lower module, around monitoring
The sensing module of travel condition of vehicle and shape;For the processing module calculated according to sensing module monitoring information;For
Obtained a result according to processing module the alarm module warned;For being carried to sensing module, processing module and alarm module
For the power module of supply of electric power;
The sensing module includes being provided with timing transmitting element in camera and laser range finder, sensing module;
The processing module includes, and handles single for carrying out the image borderization of marginalisation processing to surrounding vehicles image
Member;Areal calculation unit for calculating unit interval inward flange image area accounting variation tendency;
By with rate calculation unit synchronously draw the spacing of body vehicle and surrounding vehicles relative distance variation tendency based on
Calculate unit;For areal calculation unit and distance computation unit to be drawn into data calculate in the surrounding vehicles unit interval
The rate calculation unit of movement rate;Show that the RATES of relative motion state is single for carrying out contrast with body car speed
Member;The dangerous budget list of budget is carried out to danger coefficient for being obtained a result according to rate calculation unit and RATES's unit
Member;Memory cell for storing vehicle and all kinds of vehicle information;In picture and memory cell after for handling marginalisation
The Model Matching unit that the progress of storage information is matched.
Alarm module is to be provided with to send dangerous budget unit result to the voice of alarm module in loudspeaker, processing module
Driver element.
The present apparatus is arranged on automobile, and some sensing modules are correspondingly arranged at motor vehicle environment.Laser range finder includes laser
Transmitting terminal, laser pick-off end.
The image partition holding of the monitoring image for sensing module is provided with memory cell;It is provided with and is directed in processing module
The timing cleaning unit of image partition holding.
Timing transmitting element timing the range information monitored and image information are transmitted, when find surrounding vehicles and
Body vehicle distances it is small to a fixed value when, monitor and transmit in real time.
The model of all angles for the basic vehicle that is stored with memory cell, including two wheels, three-wheel, the headstock of four-wheel, car
Tail, strabismus model etc.;In addition, each type of vehicle performance parameter of correspondence, such as power property arguments (highest car are included
Speed, braking time etc.), structural parameters (overall width, vehicle commander etc.).
Sensing module, which is disposed in the interior, can see the place of surrounding clearly, or sensing module is located at the outside surrounding of car, typically
Be the car taken shelter from rain under, installation region be the blind spot of human eye, such as car after, the car left and right sides etc..
It is simple in construction, the rate value of the car is drawn by image change trend and the relative distance change to surrounding vehicles,
The speed of body vehicle is introduced, by the direction of relative movement for calculating surrounding vehicles;By the surrounding vehicles monitored with depositing
The performance for determining the surrounding vehicles is compared in the various information of vehicles stored in storage unit, and whether the car is drawn by computing
It is dangerous to body vehicle, information warning is played out by alarm module, effectively reminds driver to note eliminating driving blind area car
, it is to avoid occur unexpected traffic accident.
Image partition holding is set, image or video recording can be preserved, make the present apparatus that there is the work of running distance instrument
With;Timing cleaning unit is set, useless video or image is regularly cleared up, reduces memory usage.
A kind of image procossing real-time early warning system of embodiment three, referring to Fig. 1-3:Including with lower module, around monitoring
The sensing module of travel condition of vehicle and shape;
For the processing module calculated according to sensing module monitoring information;
The alarm module warned for being obtained a result according to processing module;
Power module for providing sensing module, processing module and alarm module supply of electric power;
Sensing module includes being provided with timing transmitting element in camera and laser range finder, sensing module;
Processing module includes,
Image border processing unit for carrying out marginalisation processing to surrounding vehicles image;
Areal calculation unit for calculating unit interval inward flange image area accounting variation tendency;
By with rate calculation unit synchronously draw the spacing of body vehicle and surrounding vehicles relative distance variation tendency based on
Calculate unit;
For areal calculation unit and distance computation unit to be drawn into data carry out calculating the surrounding vehicles unit interval
The rate calculation unit of interior movement rate;
For carrying out RATES's unit that contrast draws relative motion state with body car speed;
The danger for carrying out budget to danger coefficient for being obtained a result according to rate calculation unit and RATES's unit is pre-
Calculate unit;
Memory cell for storing vehicle and all kinds of vehicle information;
The Model Matching unit that picture after for handling marginalisation is matched with the progress that memory cell stores up information.
Alarm module is loudspeaker, is provided with the processing module and sends dangerous budget unit result to alarm module
Voice driven unit.
The present apparatus is arranged on automobile, and some sensing modules are correspondingly arranged at motor vehicle environment.The laser range finder
Including Laser emission end, laser pick-off end.
The image partition holding of the monitoring image for sensing module is provided with memory cell;It is provided with the processing module
Unit is cleared up in timing for image partition holding.
A kind of image procossing real time early warning method, including step once,
A. surrounding vehicles are monitored by sensing module, and the image and range information that sensing module is monitored
Timing transmission is to processing module;
B. image border processing unit carries out marginalisation processing, unit of account time inner edge to the image information received
The variation tendency of edge image relative total area accounting size, and according to sensing module Monitoring Data synchronously draw body vehicle with
The relative spacing variation tendency of surrounding vehicles;
C. set in the database of storage vehicle and all kinds of vehicle speed, the figure and database that marginalisation image is drawn
Data carry out contrast and draw the affiliated type of surrounding vehicles;
D. variation tendency and spacing variation tendency are accounted for according to area and fortune in the surrounding vehicles unit interval is drawn by computing
Dynamic speed;
E. body vehicle movement is imported, with reference to the step d bodies drawn and the relative fortune in the surrounding vehicles unit interval
Dynamic information, carries out danger coefficient according to relative motion relation and the result that may occur is moved according to current state to carry out budget;
F. danger coefficient and the result that may occur are sent to alarm module.
The image partition holding of the monitoring image for sensing module is provided with database;To image partition holding memory storage
Picture be timed cleaning.
This method is used for vehicle early warning, the alarm module is loudspeaker.
Timing transmitting element timing the range information monitored and image information are transmitted, when find surrounding vehicles and
Body vehicle distances it is small to a fixed value when, uninterruptedly monitor and transmit.
The model of all angles for the basic vehicle that is stored with memory cell, including two wheels, three-wheel, the headstock of four-wheel, car
Tail, strabismus model etc.;In addition, each type of vehicle performance parameter of correspondence, such as power property arguments (highest car are included
Speed, braking time etc.), structural parameters (overall width, vehicle commander etc.).
Sensing module, which is disposed in the interior, can see the place of surrounding clearly, or sensing module is located at the outside surrounding of car, typically
Be the car taken shelter from rain under, installation region be the blind spot of human eye, such as car after, the car left and right sides etc..
Image border Processing Algorithm is as follows:Wherein described image marginalisation processing unit is carried out to the image of image information
Rim detection and extraction;
Rim detection and extraction comprise the following steps:Make to be filtered image f (x, y) using Gaussian function first to obtain
One smoothed data array:
S (x, y)=f (x, y) * G (x, y, σ)
Wherein σ is the stroll parameter of Gaussian function, and it reflects smoothness.
Secondly, gradient calculation is carried out, smoothed data array S (x, y) gradient can be approximate with 2 × 2 first differences point
Formula calculates two array P (x, y) and the Q (x, y) of x and y partial derivatives.
The average of finite difference is sought in 2 × 2 squares, the partial derivative gradient of same point x and y in the picture is calculated,
Amplitude and azimuth are shown below respectively.
θ (x, y)=arctan (Q (x, y)/P (x, y))
In formula, arctan function contains two parameters, and result of calculation is an angle, and value mode is circumference range.
, can be by edge direction with 45 ° of divisions, eight directions, by gradient side when being aware of after the direction θ (x, y) of gradient
To the adjacent pixels in this pixel gradient direction can be found.
In order to ensure edge accurate positioning, and single-phase is kept to answer, to the ridge band in gradient magnitude image M (x, y)
Refined, retain the maximum point of amplitude localized variation, during non-maximum restraining, for each point m (x, y) of image
And the gradient direction obtained in previous step, it is compared with 2 values on gradient direction, if m (x, y) is less than gradient
2 gradient magnitudes on direction, it is not local maximum to illustrate the point, then the corresponding edges of m (x, y) is expressed as into 0, i.e. the point
It is not marginal point.
Low threshold Lth and high threshold Hth that dual threshold method detection and connection edge need are set, gradient image is carried out
Dual threashold value is handled, and what gradient magnitude was more than high threshold Hth is edge, and what gradient magnitude was less than Low threshold Lth is not edge, ladder
Degree amplitude falls between, and judges to whether there is the edge pixel more than high threshold Hth in eight field pixels of the pixel,
If in the presence of if it be edge pixel, be not otherwise to determine after a marginal point, continuous search and track edge, marginal point is connected
Afterwards;By the edge extracting of whole image, the image in image information is divided by the marginalisation of extraction, object is realized
With the segmentation of background.
The specific algorithm of Model Matching unit is:Vehicle can be obtained after treatment to vehicle image to external boundary, display
The shape of vehicle, in a closed curve.N number of sampled point is taken by equal intervals to a closed curve, on x/y plane from
Arbitrfary point (x0, y0) starts, and obtains N number of boundary point (x0, y0) in the counterclockwise direction, (x1, y1), (x2, y2) ..., (xk,
Yk), k=0,1,2 ..., N-1, s (k)=xk+jyk is expressed as with complex variable, and the coordinate sequence is periodic function, with curve
The Zhou Changwei cycles.Fast discrete Fourier transformation FFT is carried out to coordinate sequence in a cycle:
Complex coefficient F (u) is the Fourier descriptor on border, and the array describes the change on border.When the curve rotates θ
(Δ x, Δ y) and r times of amplification, FFT border starting point is moved after a sampled point along closed curve, multiple system for angle, translation
Number is:
F ' (u)=FFT [(x '+y ') rejθ+(Δx+iΔy)]
=rejθFFT(x′+y′)+FFT(Δx+iΔy)
=reiθ·e-j2πk/N s(k)+FFT(Δx+iΔy)
In formula, xk′+iyk'=xk+a+iyk+a, the motion of vehicle and the visual field of camera can cause above-mentioned curve translation,
Amplification and the change of FFT starting points, it is therefore desirable to extract curve not with the parameter of above-mentioned factor.It has been observed that working as curvilinear translation
When, (Δ x+i Δ y) values change the only FFT of k=0 components.K=0 is low frequency coefficient, corresponding to u=N/2.Therefore every
The amplitude divided by F (N/2) of one complex coefficient (except u=N/2), obtain unitary Fourier descriptor d (u):
D (u)=| | F (u) | |/| | F (N/2) | |, u=0,1,2 ..., N-1
Fourier descriptor d (u) after normalization can calculate the similarity degree between any 2 vehicles i and j, be defined as
Dispersion:
When de level off to 0 when, represent two cars shape it is basically identical;De is bigger, illustrates that the difference of vehicle shape is bigger.
Area ratio shared in whole image area y the area x of the model in the unit of account time per two field picture, enters
And count the area in the unit interval and, than f (x, y), analyze size variation trend f ' (x, y) of the area than f (x, y).
Vehicle distance data g (z) in the statistical unit time, analyzes the size variation trend g ' (z) of the vehicle distance data.
Compare the area than size variation trend f ' (x, y) and the size variation of the vehicle distance data become g ' (z), if both
Differed within the unit interval within 0~10%, then appoint and take a value, try to achieve the rate value V within the unit interval.If both are in list
Differed by more than in the time of position within 10%, take both average value V=[f ' (x, y)+g ' (z)]/2, try to achieve within the unit interval
Rate value V;
Velocity amplitude f (v) of this car within the unit interval is imported, if V values are more than 1, then it represents that other side is close, and speed
It is worth for V1=f (v)+V, depending on position of the direction by specific sensing module arrangement.If V values are less than 1, then it represents that other side to be remote,
And speed is worth for V2=f (v)-V, depending on position of the direction by specific sensing module arrangement.If V values are equal to 1, then it represents that B cars and A
Vehicle speed is identical.
The performance parameter of the successful vehicle of binding model matching unit contrast and speed, the row of the car of the car monitored
Direction is sailed, calculating obtains braking distance, the braking distance normal spacing corresponding with the speed is contrasted, and calculating is pressed
How long travel at this rate following interior may occur the data results such as accident.The result of calculation that would be possible to occur accident,
Progress voice broadcast early warning is combined with the motion state of this car.Result of calculation includes:Type of vehicle and this car travel direction, car
Speed, the estimated accident that may occur by the car motion state.
Between the speed (speed, travel direction) of surrounding vehicles, vehicle is determined, two cars are obtained with laser range finder
After spacing, the performance parameter that processing module combines the car is analyzed:
1. cart, if the speed of other side is in 10km/h-30km/h, for ordinary bicycle;If the speed of other side exists
More than 30km/h then thinks motorcycle.If travel direction is remote, it need not calculate;If travel direction is close, according to V
Value, braking distance is obtained according to the performance of B cars.Compare braking distance and now both spacing S, if braking distance is less than or waited
In spacing S, then alarm;If braking distance is not alarmed more than spacing S.
2. tricycle, if the speed of other side is in 10km/h-20km/h, for pedicab;If the speed of other side exists
More than 20km/h then thinks motor tricycle.If travel direction is remote, it need not calculate;If travel direction is close, root
According to V values, braking distance is obtained according to the performance of surrounding vehicles.Compare braking distance and now both spacing S, if braking distance
Less than or equal to spacing S, then alarm;If braking distance is not alarmed more than spacing S.
3. carriage, if travel direction is remote, need not be calculated;If travel direction is close, according to V values, according to B
The performance of car obtains braking distance.Compare braking distance and now both spacing S, if braking distance is less than or equal to spacing S,
Then alarm;If braking distance is not alarmed more than spacing S.
Carriage in 3rd point can be divided into:
1st, minicar (A00)
(1) vehicle commander is less than 3.7M;
(2) wheelbase is less than:2.35M;
(3) representative vehicle:QQ、BYD F0;
2nd, compact car (A0)
(1) vehicle commander is less than 4.3M;
(2) wheelbase is less than:2.5M;
(3) representative vehicle:206、POLO;
3rd, compact sedan (A)
(1) vehicle commander is less than 4.6M;
(2) wheelbase is less than:2.7M;
(3) representative vehicle:FOCUS、SAGITAR、307;
4th, in-between car (B)
(1) vehicle commander is less than 4.9M;
(2) wheelbase is less than:2.8M;
(3) representative vehicle:PASSAT、MAGTON、Audi A4;
5th, medium-and-large-sized car (C)
(1) vehicle commander is less than 5.1M;
(2) wheelbase is less than:2.9M;
(3) representative vehicle:AUDI A6、BMW 5;
6th, luxurious car (D)
(1) vehicle commander is more than 5.1M;
(2) wheelbase is more than:2.9M;
(3) representative vehicle:AUDI A8、BMW 7;
7. truck etc..
The inertia coeffeicent of the vehicle, the travel speed V of the car are obtained according to different automobile types and is worth to braking distance, are compared and are stopped
Car distance and now both spacing S, if braking distance is less than or equal to spacing S, alarm;If braking distance is more than spacing S
Do not alarm.
In the above-described embodiments, if surrounding vehicles are travelled with body vehicle in parallel lines, it is divided into 2 kinds of situations:One
It is front and rear traveling, two be not in the ridden in left or right direction of same straight line.Second of situation does not produce danger, if during front and rear traveling being phase
Opposite direction does not also produce danger.
If surrounding vehicles and body vehicle are travelled not in parallel lines, travel direction a, the angle with an angle
It can be measured by laser range finder, then surrounding vehicles are travelled in a angles by given pace, then the surrounding vehicles rate value calculated
It is to be travelled by a angles.If travelling in the opposite direction, danger is not also produced.
Claims (10)
1. a kind of image procossing real-time early warning system, it is characterised in that including with lower module,
Sensing module for monitoring surrounding vehicles running status and shape;
For the processing module calculated according to sensing module monitoring information;
The alarm module warned for being obtained a result according to processing module;
Power module for providing sensing module, processing module and alarm module supply of electric power;
The sensing module includes being provided with timing transmitting element in camera and laser range finder, sensing module;
The processing module includes,
Image border processing unit for carrying out marginalisation processing to surrounding vehicles image;
Memory cell for storing vehicle and all kinds of vehicle information;
The Model Matching unit that picture after for handling marginalisation is matched with the progress that memory cell stores up information;
Areal calculation unit for calculating unit interval inward flange image area accounting variation tendency;
For with rate calculation unit synchronously draw the distance computation list of body vehicle and surrounding vehicles relative distance variation tendency
Member;
Transported for areal calculation unit and distance computation unit to be drawn into data calculate in the surrounding vehicles unit interval
The rate calculation unit of dynamic speed;
For carrying out RATES's unit that contrast draws relative motion state with body car speed;
The dangerous budget list of budget is carried out to danger coefficient for being obtained a result according to rate calculation unit and RATES's unit
Member.
2. image procossing real-time early warning system according to claim 1, it is characterised in that the alarm module is to raise one's voice
It is provided with device, the processing module and sends dangerous budget unit result to the voice driven unit of alarm module.
3. image procossing real-time early warning system according to claim 1, it is characterised in that the system is arranged on automobile,
Some sensing modules are correspondingly arranged at motor vehicle environment.
4. image procossing real-time early warning system according to claim 1, it is characterised in that the laser range finder includes swashing
Light transmitting terminal, laser pick-off end.
5. image procossing real-time early warning system according to claim 1, it is characterised in that pin is provided with the memory cell
To the image partition holding of the monitoring image of sensing module.
6. image procossing real-time early warning system according to claim 5, it is characterised in that pin is provided with the processing module
Unit is cleared up in timing to image partition holding.
7. a kind of image procossing real time early warning method, it is characterised in that including step once,
A. surrounding vehicles are monitored by sensing module, and image and the range information timing that sensing module is monitored
Transmit to processing module;
B. image border processing unit carries out marginalisation processing, unit of account time inward flange to the image information received
The variation tendency of image relative image gross area accounting size, and according to sensing module Monitoring Data synchronously draw body vehicle with
The relative spacing variation tendency of surrounding vehicles;
C., data in the database of storage vehicle and all kinds of vehicle information, the figure and database that marginalisation image is drawn are set
Carry out contrast and draw the affiliated type of surrounding vehicles;
D. the motion in the surrounding vehicles unit interval is drawn by computing according to area accounting variation tendency and spacing variation tendency
Speed;
E. body state of motion of vehicle information is imported, with reference to the step d bodies drawn and the phase in the surrounding vehicles unit interval
To movable information, the result that carries out danger coefficient according to relative motion relation and may occur according to current motion state is carried out in advance
Calculate;
F. danger coefficient and the result that may occur are sent to alarm module.
8. image procossing real time early warning method according to claim 7, it is characterised in that be provided with the database for sense
Survey the image partition holding of the monitoring image of module.
9. image procossing real time early warning method according to claim 8, it is characterised in that to described image partition holding internal memory
The picture of storage is timed cleaning.
10. image procossing real time early warning method according to claim 7, it is characterised in that this method is used for vehicle early warning,
The alarm module is loudspeaker.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710228535.8A CN107226023A (en) | 2017-04-10 | 2017-04-10 | Image procossing real-time early warning system and method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710228535.8A CN107226023A (en) | 2017-04-10 | 2017-04-10 | Image procossing real-time early warning system and method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107226023A true CN107226023A (en) | 2017-10-03 |
Family
ID=59933687
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710228535.8A Pending CN107226023A (en) | 2017-04-10 | 2017-04-10 | Image procossing real-time early warning system and method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107226023A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110728844A (en) * | 2019-09-11 | 2020-01-24 | 平安科技(深圳)有限公司 | Traffic light self-adaptive control method and device, traffic control equipment and storage medium |
CN113450574A (en) * | 2018-07-13 | 2021-09-28 | 神讯电脑(昆山)有限公司 | Estimation method and device for moving speed of vehicle and monitoring system |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2000074645A (en) * | 1998-08-27 | 2000-03-14 | Yazaki Corp | Device and method for monitoring periphery |
CN1457023A (en) * | 2002-05-09 | 2003-11-19 | 松下电器产业株式会社 | Monitor, monitoring method and programm |
CN103204123A (en) * | 2013-03-25 | 2013-07-17 | 中国电子科技集团公司第三十八研究所 | Vehicle-pedestrian detecting, tracking and early-warning device and method |
CN103593981A (en) * | 2013-01-18 | 2014-02-19 | 西安通瑞新材料开发有限公司 | Vehicle model identification method based on video |
CN104129387A (en) * | 2014-07-25 | 2014-11-05 | 杭州电子科技大学 | Single-camera automobile collision avoidance method based on risk balancing of safe distance and collision time |
-
2017
- 2017-04-10 CN CN201710228535.8A patent/CN107226023A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2000074645A (en) * | 1998-08-27 | 2000-03-14 | Yazaki Corp | Device and method for monitoring periphery |
CN1457023A (en) * | 2002-05-09 | 2003-11-19 | 松下电器产业株式会社 | Monitor, monitoring method and programm |
CN103593981A (en) * | 2013-01-18 | 2014-02-19 | 西安通瑞新材料开发有限公司 | Vehicle model identification method based on video |
CN103204123A (en) * | 2013-03-25 | 2013-07-17 | 中国电子科技集团公司第三十八研究所 | Vehicle-pedestrian detecting, tracking and early-warning device and method |
CN104129387A (en) * | 2014-07-25 | 2014-11-05 | 杭州电子科技大学 | Single-camera automobile collision avoidance method based on risk balancing of safe distance and collision time |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113450574A (en) * | 2018-07-13 | 2021-09-28 | 神讯电脑(昆山)有限公司 | Estimation method and device for moving speed of vehicle and monitoring system |
CN110728844A (en) * | 2019-09-11 | 2020-01-24 | 平安科技(深圳)有限公司 | Traffic light self-adaptive control method and device, traffic control equipment and storage medium |
CN110728844B (en) * | 2019-09-11 | 2022-02-22 | 平安科技(深圳)有限公司 | Traffic light self-adaptive control method and device, traffic control equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10203697B2 (en) | Real-time image-based vehicle detection based on a multi-stage classification | |
US9477893B2 (en) | Positioning system and method for positioning a vehicle | |
CN104180818B (en) | A kind of monocular vision mileage calculation device | |
CN108437893B (en) | Early warning method and device for automobile deviating from lane | |
CN108749819B (en) | Tire vertical force estimating system and evaluation method based on binocular vision | |
CN103927754A (en) | Calibration method of on-board camera | |
US20150266509A1 (en) | Object detection during vehicle parking | |
EP3009789A1 (en) | Monocular-motion-stereo distance estimation method, and monocular-motion-stereo distance estimation apparatus | |
CN103413325B (en) | A kind of speed of a motor vehicle authentication method based on vehicle body positioning feature point | |
CN102915532A (en) | Method of determining extrinsic parameters of a vehicle vision system and vehicle vision system | |
CN105151047A (en) | Automobile gravity center slip angle measuring method | |
CN105930787A (en) | Vehicle door opening early-warning method | |
CN103909933A (en) | Method for estimating lateral force of front wheels of distributed-type electrically-driven vehicle | |
CN101941438A (en) | Intelligent detection control device and method of safe interval | |
CN110341714B (en) | Method for simultaneously estimating vehicle mass center slip angle and disturbance | |
EP2325607A1 (en) | Vehicle estimate navigation appartus, vehicle estimate navigation method, and vehicle estimate navigation program | |
CN110930323A (en) | Method and device for removing light reflection of image | |
CN103686083A (en) | Real-time speed measurement method based on vehicle-mounted sensor video streaming matching | |
CN107226023A (en) | Image procossing real-time early warning system and method | |
CN103473948A (en) | Piezoelectric cable layout structure and overloaded vehicle wheel line horizontal position identification method | |
CN111006884B (en) | Method for measuring wheel axle slip angle and slip stiffness based on Fourier transform | |
CN109131336A (en) | Obtain the method and system of coefficient of road adhesion | |
Wang et al. | Vision-based vehicle body slip angle estimation with multi-rate Kalman filter considering time delay | |
CN106197471A (en) | Road Detection spacing based on information fusion triggers device and triggering method | |
CN103568990A (en) | Method and system for achieving vehicle safety warning |
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 |
Application publication date: 20171003 |
|
RJ01 | Rejection of invention patent application after publication |