CN103065470B - Detection device for behaviors of running red light of vehicle based on machine vision with single eye and multiple detection faces - Google Patents
Detection device for behaviors of running red light of vehicle based on machine vision with single eye and multiple detection faces Download PDFInfo
- Publication number
- CN103065470B CN103065470B CN201210551907.8A CN201210551907A CN103065470B CN 103065470 B CN103065470 B CN 103065470B CN 201210551907 A CN201210551907 A CN 201210551907A CN 103065470 B CN103065470 B CN 103065470B
- Authority
- CN
- China
- Prior art keywords
- vehicle
- image
- signal lamp
- track
- red light
- 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.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 79
- 230000006399 behavior Effects 0.000 title abstract description 21
- 230000033228 biological regulation Effects 0.000 claims abstract description 65
- 238000000034 method Methods 0.000 claims abstract description 51
- 230000033001 locomotion Effects 0.000 claims abstract description 40
- 238000012544 monitoring process Methods 0.000 claims abstract description 23
- 238000012545 processing Methods 0.000 claims abstract description 8
- 230000008569 process Effects 0.000 claims description 25
- 230000003542 behavioural effect Effects 0.000 claims description 24
- 238000003384 imaging method Methods 0.000 claims description 13
- 230000000694 effects Effects 0.000 claims description 9
- 230000005540 biological transmission Effects 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 239000003086 colorant Substances 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 238000009434 installation Methods 0.000 claims description 6
- 238000012790 confirmation Methods 0.000 claims description 5
- 230000009471 action Effects 0.000 claims description 4
- 238000013461 design Methods 0.000 claims description 4
- 238000005286 illumination Methods 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 239000000463 material Substances 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 4
- 230000001939 inductive effect Effects 0.000 description 4
- 230000001815 facial effect Effects 0.000 description 3
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000006698 induction Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 206010015946 Eye irritation Diseases 0.000 description 1
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 231100000013 eye irritation Toxicity 0.000 description 1
- 239000000945 filler Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Landscapes
- Traffic Control Systems (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
A detection device for behaviors of running the red light of a vehicle based on machine vision with a single eye and multiple detection faces comprises a microprocessor and a monitoring sensor, wherein the monitoring sensor is used for monitoring road conditions. The monitoring sensor is a high definition camera with the single eye and the multiple detection faces. The high definition camera with the single eye and the multiple detection faces is used for shooting signal lamps and an image relative to the signal lamps and of the vehicle running the red light near a stop mark of a road. An image processing technique is used for identification of the states of the traffic signal lamps and the motion states of the vehicle near the stop mark. The behaviors of running the red light of the vehicle are automatically judged according to relevant laws and regulations of the state, and illegal parking recordings and relevant evidence materials are automatically generated. The detection device for the behaviors of running the red light of the vehicle based on the machine vision with the single eye and the multiple detection faces is accurate in detection, convenient to maintain and install, and complete in evidence obtaining of rule and regulation breaking behaviors.
Description
Technical field
The present invention relates to a kind of running red light for vehicle behavioral value device, effective to machine vision technique, image processing techniques, data communication technology, computer networking technology etc. integrated use is managed in break in traffic rules and regulations, is particularly useful for special detection and the electronic police processing behavior of making a dash across the red light.
Background technology
Along with the develop rapidly of economy, the sharply increase of vehicle fleet size, create day by day serious traffic problems, as traffic environment deterioration, frequent accidents, congested in traffic obstruction etc., particularly in Beijing, Shanghai etc., medium-and-large-sized urban transportation situation presents degradating trend, the death toll of the traffic hazard of annual China is in 100,000 people's left and right horizontal substantially, occupies first place in the world for successive years.In China, traffic hazard is dead 3 people often, have 2 to be due to the traffic violation.China's statistics of traffic accidents data display for many years, the main inflicter of traffic hazard is automobile driver, and pedestrian, rider and people are by bike three in traffic hazards is injured greatly colony, 3/4ths in death toll is pedestrian, rider and people by bike.
Many statistical data show, the serious traffic hazard of thumping majority occurs in crossroad, and thumping majority wherein breaks rules and regulations due to automobile driver or make a dash across the red light to cause.Video camera has generally been used to monitor the violating the regulations of traffic in current China in metropolitan traffic administration, owing to lacking the support of image processing techniques and artificial intelligence technology, also need a large amount of manpowers to carry out real-time observation, thus real-time reflection is made to abnormal surface conditions, this disposal route needs a large amount of human inputs, requires again higher quality and notice to monitoring personnel simultaneously.Adopt the shortcoming of this working method be take a large amount of police strength, law enfrocement official easily tired, can not all weather operations, cannot strong evidence be provided.
A kind of running red light for vehicle behavioral value device based on monocular many detection faces machine vision of the present invention is used for the traffic intersection of unattended duty, device can be monitored the situation of traffic intersection automatically, the travel conditions of vehicle is traced and monitored, automatically records the vehicle violated the traffic regulations; In invention the crucial problem that will solve be realize can correctly, the moving vehicle that makes a dash across the red light in quick detection and Identification video sequence, obtained video information can be used for law ruling like this.
The prevailing traffic rule of China is that State Council issued on March 9th, 1988, has clear and definite regulation in rule to crossing vehicle, the current of pedestrian.Specify in chapter 2 Article 10 of rule: during (1) green light, permit vehicle, walk, but turn vehicle forbidden hinder keep straight on vehicle and by the walk of letting pass; (2) when amber light is bright, inaccurate vehicles and pedestrians, but crossed the vehicle of stop line and entered the pedestrian of crossing, can continue to pass through; (3) during when red, inaccurate vehicle, walk; (4), when green arrow lamp is bright, permits vehicle and pass through in a direction indicated by the arrow; (5), during amber light flicker, vehicle, pedestrian must guarantee to pass through under safe principle.Without the through vehicles of lateral road on the right of the vehicle of turning right and T-shaped crossing, run into preceding article (2), (3) rules and regulations regularly, when do not hinder by let pass vehicles and pedestrians, P Passable.
The existing traffic law of China can be concluded with table 1 in the relation of signal lamp and traveling state of vehicle;
| Red light | Amber light | Green light | Vehicle condition |
| Bright | Go out | Go out | Stop |
| Bright | Bright | Go out | Preparation is advanced |
| Go out | Bright | Go out | Prepare to stop |
| Go out | Go out | Bright | Advance |
The relation of table 1 signal lamp and traveling state of vehicle
Made a dash across the red light this traffic violation to stop and brought the phenomenon of threat to people's life in the past, the effect playing restricted mechanism on duty of traffic-police can only be utilized; Now due to the attention of traffic department, both at home and abroad by image, camera, sensor high-technology in traffic intersection, these effort cause the type systematic being called red light video camera (RLC) to produce, and it often works together with traffic grade circuit and the inductive coil monitor under being embedded in road surface.
Detection technique of making a dash across the red light starts from late 1980s, and since 1994, the industrially developed country such as the U.S., France succeeds in developing the system of making a dash across the red light in succession.The surveillance of making a dash across the red light of France's development adopts the mode of spark photograph, and the ground such as san francisco, usa use geomagnetic induction coil in conjunction with CCD acquisition mode.China introduced the surveillance of making a dash across the red light of photographic means in 1996, and the units such as Beijing Communication scientific research institution have developed black and white type red light surveillance.Shenzhen Green's prestige transport science and techonologies company limited in 1997 have developed the CCD colour detected based on geomagnetic induction coil and to make a dash across the red light surveillance, and Northwestern Polytechnical University then have developed the colored CCD detected based on video and to make a dash across the red light surveillance.The intelligent detecting method that makes a dash across the red light at present based on image processing techniques mainly contains two kinds: (1) inductive coil monitor combines the detection method of making a dash across the red light of taking a picture or making a video recording; (2) completely based on the detection method of video technique.
In use in the vehicle peccancy of traffic intersection, although also there are the imagination of this respect in many places and scientific research institution, also automatic detection and indentification technology is not applied to the practical application at crossing so far, just the method that may use is set forth.At the traffic intersection in flourishing city, what usually adopt is the mode of operation of camera-sensing coil, under coil is embedded in earth's surface, by the technical controlling such as vibration-sensing, hot sensing camera shooting picture violating the regulations.This mode has following several significant shortcoming: the cost of (1) system is high.Because whenever the monitoring point that increase by is new, human and material resources just must be dropped into and financial resources bury sensing coil underground at traffic intersection; (2) crossing to be monitored is dumb.Because can only monitor at the crossing of to have buried coil; (3) maintenance of system is inconvenient, if because inductive coil damages, need again to bury, lay; (4) affecting the normal operation of traffic, because when burying inductive coil, needing the traffic blocking road surface.
The state of signal lamp is the foundation judging whether vehicle breaks rules and regulations.When monitoring vehicles peccancy, the state of signal lamp can be obtained by Access Control line sometimes.But, surveillance can not be accessed easily at the control line of some local traffic lamp, desirably by the method for image procossing, the status information of signal lamp can be obtained, the intelligent of system and dirigibility can be improved like this.
On the other hand, the white bar on crossing is used to remind driver and pedestrian, if when the state of signal lamp is red light, vehicle or pedestrian cross this white bar (stop line) and be just considered to make a dash across the red light.In image procossing, in order to identify whether vehicle surmounts stop line, just need the position determining stop line.And the state of marker lamp with identify whether vehicle surmounts stop line and must carry out simultaneously.The running red light for vehicle behavioral value device based on monocular many detection faces machine vision so just can be made to have real-time, accurate, intelligentized feature.
But, from visual angle on road, the state of signal lamp and the optimal viewing angle of stop line are on direction of traffic, this technically also exists some difficulties at present to the image information that will obtain with a video camera near the state of signal lamp and stop line simultaneously, although two cameras can be adopted, a state for detection signal lamp, another is for detecting the state of vehicle near stop line and pedestrian.Although this mode is feasible, the complexity of cost and system can be increased, the real-time that impact detects.
The Ministry of Public Security revises " motor vehicle driving license is applied and used regulation " of issuing, and came into effect from 1 day January in 2013, make a dash across the red light 6 points, button, is based on traffic police's on-site law-enforcing, and it is not yet clear and definite that current electronic police exposes." situations of 200 yuan of punishment 6 points, buttons are based on scene punishment; for electric police grasp shoot; temporarily will not score at present; because current motor vehicle is as walking-replacing tool; not every motor vehicle is all driven by car owner; so be difficult to determine the phenomenon of making a dash across the red light of electric police grasp shoot is difficult to identify which driver, so temporarily will not deduct points at present to these a part of illegal activities of electric police grasp shoot.But " along with the widespread use of high-definition camera, as long as during the facial image of the electronic police driver that can significantly capture, electronic police carrys out alternative on-site law-enforcing and just comes within a measurable distance.
National inventing patent numbers 200610053154.2 discloses a kind of monitoring vehicle breaking regulation method and apparatus based on omnidirectional computer vision, this invention adopts a kind of omnibearing vision sensor to monitor vehicle-state and the signal lamp state at whole crossing, thus has judged whether break in traffic rules and regulations behaviors such as making a dash across the red light.Omnibearing vision sensor has a panorama to obtain the advantage of whole 360 ° of physical environment video images, is arranged on the act of violating regulations that all phase places in crossing can be detected in the middle of crossing; But also there is obvious defect simultaneously, owing to having carried out the physical compression of image through the image of omnibearing vision sensor, make the image of traffic lights be difficult to identify accurately; On the other hand, because compression of images amount causes greatly, very large difficulty is produced to the license plate number identification of vehicles peccancy and the confirmation of driver's face.Although speed dome camera and omnibearing vision sensor can be adopted to merge, capture the image in license plate number and region, driver's seat with quick, such solution will certainly increase the cost of system and the complexity of control.Along with the fast development of high-definition camera, identify to the behavior of making a dash across the red light and car plate simultaneously and bring possibility.
National inventing patent numbers 201210239881.3 discloses a kind of motor vehicle red light violation evidence collecting method based on HD video, and adopt high-definition camera capture video, the object detecting and tracking technology according to video content is analyzed vehicle action trail.Under the red light phase place of traffic lights, the behavior of making a dash across the red light is detected, effective high-definition image violating the regulations is obtained after confirming as red light violation, formation meets two evidence obtaining pictures that make a dash across the red light of national standard (GA/T832-2009), and will record the video flowing synthesis evidence obtaining video of whole red light violation process.Finally by network, evidence obtaining result is delivered to the management system of law enforcement agency.The greatest problem of this invention: high-definition camera is arranged on after vehicle, the sight line of stop line is easily by occlusion, easily cause the error of line ball and more line judgement, and directly can not photograph the frontal one image of driver, also need the phase place accessing signal lamp from teleseme in addition.National inventing patent number 201010598795.2 Jaywalking snapshot methods and device also also exist same problem.
According to " red-lamp running automatic recording system general technical specifications " (GA/T496-2004), complete the video capture process to red light violation behavior.Requirement in " red-lamp running automatic recording system general technical specifications " (GA/T496-2004), the behavior automatic production record of making a dash across the red light answers logging machine motor-car to make a dash across the red light the image information of two to three positions in process to reflect that automobile driver makes a dash across the red light illegal activities process, the information of first position clearly should be able to distinguish that motor vehicle is pressed in or crosses the situation of stop line, second and the image information of the 3rd position should be able to reflect that whole vehicle body has crossed the situation of stop line.
The present invention is devoted to solve following problem: adopt a high-definition camera can photograph the associated picture of the vehicles running red light near the signal lamp track stop line relevant with signal lamp simultaneously, namely the image of first position, second position and the 3rd position, also needs the license plate image information comprising discernible driver's facial image information and steering vehicle in captured image; Automatic identification and judge red light violation behavior, automatically identify the car plate of vehicles peccancy, automatically enforces the law for electronic police and provides vaild evidence.
Summary of the invention
Being difficult to take into account to overcome the existing running red light for vehicle behavioral value device based on machine vision the deficiency detecting the aspects such as accuracy, maintenance and the convenience of installation, integrality of collecting evidence violating the regulations, the invention provides a kind of running red light for vehicle behavioral value device detecting precisely, safeguard the machine vision based on the many detection faces of monocular that evidence obtaining easy for installation and violating the regulations is complete.
The technical solution adopted for the present invention to solve the technical problems is:
Based on a running red light for vehicle behavioral value device for the machine vision of the many detection faces of monocular, described running red light for vehicle behavioral value device comprises microprocessor and the monitoring sensor for monitoring crossing situation, and described microprocessor comprises:
View data reads and pretreatment module, passing the video image information of coming, carrying out pre-service to video image information for reading from monitoring sensor;
Virtual detection line setting, the setting of traffic lights surveyed area, track customization, track associate customized module with signal lamp, for customize track virtual stop line, customization traffic lights surveyed area, the track customized on road, customize track and associating between control signal lamp;
Network transmission module, for the image of the process violating the regulations by vehicles peccancy, when and where information by Internet Transmission to traffic police administrative authority;
Described monitoring sensor is the high-definition camera of the many detection faces of monocular, the high-definition camera of the many detection faces of described monocular comprises in order to the catadioptric minute surface of reflected signal lamp regional extent, in order to take the high-definition camera at stop line near zone travel condition of vehicle and signal lamp state, described catadioptric minute surface is positioned at the front upper place of high-definition camera, photograph the corresponding signal lamp state in monitored crossing by the catadioptric high-definition camera function of catadioptric minute surface, the state near the stop line of monitoring crossing can also be photographed by high-definition camera on the other hand; High-definition camera imaging surface has the image information of two detection faces, and its middle and upper part is the image of signal lamp state, and bottom is the image of the crossing state corresponding with this signal lamp state; All tracks on the image pickup scope covering path of a high-definition camera, the fixing corner of adjustment catadioptric minute surface, makes the top in the image of high-definition camera imaging can photograph the image of signal lamp state;
Described microprocessor also comprises:
Color space conversion module, the image rgb color space for the signal lamp region to the traffic lights surveyed area setting described in the video image on each track is transformed into HSV color space;
Signal lamp state detection module, for the detection to the signal lamp state in the video image on each track;
Moving vehicle detection module, for the detection to the moving vehicle object in the video image on each track;
Whether vehicle peccancy judge module, also exist red light violation behavior for the vehicle detected on each track;
The license plate number identification module of vehicles peccancy, for identifying the license plate number of vehicles peccancy;
Record generation module violating the regulations, for by the whole process of vehicle peccancy behavior and same time, place, travel direction generate a record violating the regulations together with the information of the license plate number of vehicles peccancy.
Further, the high-definition camera of the many detection faces of described monocular, high-definition camera and catadioptric minute surface is adopted to carry out integrated design, the supporting seat of high-definition camera draws a catadioptric mirror support bar, one end of support bar is connected with the universal auxiliary connection of catadioptric minute surface, and the lens location of the relative high-definition camera of catadioptric minute surface can adjust at any angle; When actual installation uses, by adjusting relative position and the angle of catadioptric minute surface and high-definition camera camera lens, make the top in the image of high-definition camera imaging can photograph the image of signal lamp state; Adjustment terminates the universal auxiliary connection of finger lock to ensure that high-definition camera function obtains the state of signal lamp effectively, accurately.
Further, described virtual detection line setting, the setting of traffic lights surveyed area, track customization, track associate in customized module with signal lamp, described virtual detection line setting is for setting up the virtual detection line corresponding with the stop line on real road, and the method for customization is that the road plane image directly taken by high-definition camera draws the line coincided with stop line; Described traffic lights surveyed area setting is used for the state of detection signal lamp, because the position of traffic lights is fixed, the interference of some images can be got rid of by the setting of traffic signal light condition surveyed area, to improve the accuracy detecting traffic signal light condition, the method for customization is the direct region matched by picture and traffic lights size on catadioptric mirror image parts of images taken by high-definition camera; Described track is customized for being set in direction multilane allowing vehicle operating, in some cities to clear stipulaties on multilane left lateral, craspedodrome, right lane, lane direction setting must be consistent with actual road junction roadway situation, the method of customization is that the road plane image directly taken by high-definition camera draws the lane line consistent with actual road junction roadway situation, and lane line needs to set up associating between lane line with virtual detection line after customizing; Described track associates with signal lamp and is customized for setting up associating between track with signal lamp, namely left lateral track associates with left lateral signal lamp, Through Lane associates with craspedodrome signal lamp, right lane associates with right lateral signal lamp, owing to establishing associating between lane line with virtual detection line above, therefore just establish associating of virtual detection line and signal lamp here; Just can judge whether vehicle also exists the behavior of making a dash across the red light and get over the act of violating regulations of line traveling according to such association and traffic law.
Further again, in described signal lamp state detection module, first, by color normalized, 3 component values of HSV are all between 0 ~ l; Then the number of pixels Number of statistical color histogram tone H in the green 3 kinds of color gamuts of reddish yellow is used
{ R, Y, G}, the total number of pixels of the pixel in traffic lights surveyed area is designated as Total, the ratio both calculating with formula (1),
NumRadio
{i}＝Number
{i}/Total i∈{R,Y,G} (1)
In formula, Number
{ i}for a certain color in the green 3 kinds of colors of reddish yellow is in the number of pixels of traffic lights surveyed area, Total is the total number of pixels of the pixel of traffic lights surveyed area, NumRadio
{ i}for the ratio of the total number of pixels of the pixel of a certain color in the number of pixels value and this region of traffic lights surveyed area in the green 3 kinds of colors of reddish yellow;
Finally carry out the condition adjudgement of traffic lights, determination methods is provided by formula (2),
In formula, NumRadio
{ R}for the ratio of the total number of pixels of the pixel of redness in the number of pixels value and this region of traffic lights surveyed area, NumRadio
{ Y}for the ratio of the total number of pixels of the pixel of yellow in the number of pixels value and this region of traffic lights surveyed area, NumRadio
{ G}for the ratio of the total number of pixels of the pixel of green in the number of pixels value and this region of traffic lights surveyed area.
In described moving vehicle detection module, extract the foreground moving object on this track with motion history image algorithm, judge whether there is moving vehicle object on the virtual detection line on this track;
Described motion history image algorithm, i.e. MHI algorithm, the motion history image obtained in MHI obtains after the adjacent image frame in a period of time interval carries out inter-frame difference and gray processing process, this algorithm obtaining motion history image can obtain motion outline template well for the target be kept in motion all the time and calculated amount is smaller, MHI algorithm can be used for creating a movement gradient image and the direction and the size that are calculated movement gradient by SOBEL operator, utilize this result can be further used for estimating the direction of motion of object simultaneously, just foreground moving object can be obtained because MHI algorithm does not need background modeling only to carry out inter-frame difference process, thus this algorithm has very high real-time for extraction foreground moving object, computing method are as shown in formula (3),
In formula, the binary picture sequence of D (x, y, t) moving region, τ is the duration, H
τthe motion history binary picture sequence that (x, y, t-1) was a upper moment;
Obtained the foreground moving object on track by MHI algorithm after, owing to track may also exist pedestrian or other Moving Objects, therefore need to judge whether these foreground moving object are vehicles; Foreground moving object is distinguished in the present invention by area threshold, shape and direction of motion, because the area of vehicle is larger than the area of pedestrian or cyclist, the direction of motion of pedestrian or cyclist is perpendicular to direction, track simultaneously, and shape has bigger difference, judge that computing method formula (4) represents
In formula, ∑ H
τthe sum of all pixels of the binary image that (x, y, t) is foreground object, T
1for area threshold, Shape is the shape attribute of foreground object, then finds shape attribute to obtain, T by searching border to binary image
2for rectangular shape threshold value, Direction is the direction of motion of foreground object, obtains, T by during calculating MHI algorithm
3for direction threshold value;
Effectively by the vehicle extraction that moves in track out, can then judge whether this moving vehicle is pressed in or crosses virtual detection line by the judgement of formula (4).
In described vehicle peccancy judge module, be divided into two kinds of basic conditions; The first situation has the road clearly having lane traffic lamp, and the second situation is the road not having clear and definite lane traffic lamp and only have a traffic lights;
The first situation is the road for there being clear stipulaties driveway travel directions, first vehicle lane change whether is had to travel according to virtual track is detected, if there is the vehicle on craspedodrome road to occur to turn right or turn left all to be judged to drive against traffic regulations, the situation of this situation and signal lamp has nothing to do, next is drives against traffic regulations relevant with signal lamp, this situation is will according to the situation of signal lamp, virtual track and stop the detection case in dummy line to make a decision, generally there is corresponding signal lamp for each track of each travel direction, and each travel direction also has corresponding stop line, therefore judge according to the traffic law that China is existing, be taken in certain track the when red of certain direction running time, if find that there is vehicle entering and stop dummy line or be in stopping just thinking to make a dash across the red light in dummy line position, as violating the regulations, when certain track be taken in the amber light of certain direction running bright time, entering stop dummy line as act of violating regulations if find that there is vehicle,
The second situation is the road not having clear and definite lane traffic lamp and only have a traffic lights, the judgement of this situation is more more complex than the first situation, even if this is because be allow to turn right according to the vehicle of the existing traffic law of China when red light on the right of road, that is, be not that all vehicles surmounting stopping dummy line position are all violating the regulations when red light, distinguish and whether have the result also will seeing vehicle tracking below violating the regulations; Specific practice finds that there is vehicle entering and stop dummy line or be in stopping that dummy line position just sets this vehicle and getting over line states, then continue to follow the tracks of, if follow the tracks of the vehicle finding to set more line states not bend to right, so just as act of violating regulations.
Judge to start when there is running red light for vehicle situation to record motor vehicle violation evidence obtaining video image in described vehicle peccancy judge module, start the image capturing first position; Along with passage of time, vehicle detected still towards under signal lamp disabled orientation ruuning situation, continue the image of candid photograph second and the 3rd position, the form capturing image is JPEG.
In the license plate number identification module of described vehicles peccancy, according to high-definition camera capture the vehicles peccancy image information of first position, the number on front part of vehicle car plate is identified; If the license plate number of vehicles peccancy can not be identified, continue to carry out license plate number identification by the vehicles peccancy image information of second position, when the vehicles peccancy image information of the 3rd position carry out license plate number recognition result still None-identified, allow the manual confirmation of break in traffic rules and regulations process department.
In described record generation module violating the regulations, with the specific requirement and the Index Content that include " the red-lamp running automatic recording system general technical specifications " defined specified in " the traffic safety illegal activities image forensics technical manual " and GA/T496-2004 international standard specified in GA/T832-2009 national standard in record; Capture first position, the image of second position and the 3rd position and the video image of process of making a dash across the red light all to record violating the regulations with attachment version apposition, annex name with time+mode in place names; Record violating the regulations and annex are kept in red light violation database of record.
Capture primary importance, the second place and the 3rd position red light violation behavior image time adopt a kind of narrow-pulse generator to send pulsed light carry out flash of light floor light, flash of light floor light has Microprocessor S3C44B0X, start flash of light floor light when candid photograph action simultaneously, pulsed light flash duration is 1/tens of general flash duration, flash of light floor light is arranged near monocular many detection faces video camera, and flash illumination light projects towards stop line direction, crossing.
Beneficial effect of the present invention is mainly manifested in: 1, adopt a high-definition camera can photograph the associated picture of the vehicles running red light near the signal lamp track stop line relevant with signal lamp simultaneously, detect precisely, safeguard that evidence obtaining easy for installation and violating the regulations is complete; 2, automatically identify and judge red light violation behavior, the automatic car plate identifying vehicles peccancy, peccancy detection algorithm is simple, and the ratio of performance to price is high; 3, coming into operation of this device can have very large deterrent effect to the driver of custom running red light for vehicle behavior, also greatly reduces the Spot Enforcing amount of traffic police simultaneously, improves traffic administration efficiency.
Accompanying drawing explanation
Fig. 1 is monocular many detection faces high definition camera device; 1-high-definition camera, 2-catadioptric minute surface;
Fig. 2 is the fundamental diagram of the running red light for vehicle behavioral value device of machine vision based on the many detection faces of monocular; 1-high-definition camera, 2-catadioptric minute surface, 3-traffic lights, 4-stop line, 5-track, 6-vehicle, 9-microprocessor;
Fig. 3 is the image schematic diagram of the running red light for vehicle behavioral value device of machine vision based on the many detection faces of monocular; 3-traffic lights, 4-stop line, 5-track, 6-vehicle, the car plate of 7-vehicle, the driver face of 8-vehicle;
Fig. 4 is the image that stop line-right lane control signal lamp is corresponding being divided into right lane-right lane after the imaging schematic diagram of Fig. 3 customizes by track;
Fig. 5 is the image that stop line-Through Lane control signal lamp is corresponding being divided into Through Lane-Through Lane after the imaging schematic diagram of Fig. 3 customizes by track;
Fig. 6 is the image that stop line-left-lane control signal lamp is corresponding being divided into left-lane-left-lane after the imaging schematic diagram of Fig. 3 customizes by track;
Fig. 7 is the process flow diagram that in the running red light for vehicle behavioral value device based on the machine vision of the many detection faces of monocular, some threads detect red light violation.
Embodiment
Below in conjunction with accompanying drawing, the invention will be further described.
Embodiment 1
With reference to Fig. 1 ~ Fig. 7, a kind of running red light for vehicle behavioral value device of the machine vision based on the many detection faces of monocular, based on the running red light for vehicle behavioral value device of the machine vision of the many detection faces of monocular fundamental diagram as shown in Figure 2; Described running red light for vehicle behavioral value device comprises microprocessor, for monitoring the monitoring sensor of road cross situation, described microprocessor comprises: view data reads and pretreatment module, pass the video image information of coming for reading from monitoring sensor, pre-service is carried out to video image information; Virtual detection line setting, the setting of traffic lights surveyed area, track customization, track associate customized module with signal lamp, for customize track virtual stop line, customization traffic lights surveyed area, the track customized on road, customize track and associating between control signal lamp; Network transmission module, for the image of the process violating the regulations by vehicles peccancy, when and where information by Internet Transmission to traffic police administrative authority; Described monitoring sensor is the high-definition camera of the many detection faces of monocular, the high-definition camera of the many detection faces of described monocular comprises in order to the catadioptric minute surface of reflected signal lamp regional extent, in order to take the high-definition camera at stop line near zone travel condition of vehicle and signal lamp state, described catadioptric minute surface is positioned at the front upper place of high-definition camera, photograph the corresponding signal lamp state in monitored crossing by the catadioptric high-definition camera function of catadioptric minute surface, the state near the stop line of monitoring crossing can also be photographed by high-definition camera on the other hand; High-definition camera imaging surface has the image information of two detection faces, and its middle and upper part is the image of signal lamp state, and bottom is the image of the crossing state corresponding with this signal lamp state; The high definition imager chip of 3200 × 1200 pixels more than 5,000,000 is adopted in high-definition camera, all tracks on the image pickup scope covering path of a high-definition camera, high-definition camera is arranged on the place close with signal lamp, for there being the place of signal lamp bracket, high-definition camera is placed on same support, the fixing corner of adjustment catadioptric minute surface, makes the top in the image of high-definition camera imaging can photograph the image of signal lamp state;
High-definition camera and catadioptric minute surface carry out integrated design, as shown in Figure 1, the supporting seat of high-definition camera draws a catadioptric mirror support bar, one end of support bar is connected with the universal auxiliary connection of catadioptric minute surface, and the lens location of the relative high-definition camera of catadioptric minute surface can adjust at any angle; When actual installation uses, by adjusting relative position and the angle of catadioptric minute surface and high-definition camera camera lens, make the top in the image of high-definition camera imaging can photograph the image of signal lamp state; Adjustment terminates the universal auxiliary connection of finger lock to ensure that high-definition camera function obtains the state of signal lamp effectively, accurately; On high-definition camera imaging surface, have the image information of two detection faces like this, its middle and upper part is the image of signal lamp state, and bottom is the image of the crossing state corresponding with this signal lamp state, as shown in Figure 3;
Described virtual detection line setting, the setting of traffic lights surveyed area, track customization, track associate customized module with signal lamp, described virtual detection line setting is for setting up the virtual detection line corresponding with the stop line on real road, and the method for customization is that the road plane image directly taken by high-definition camera draws the line coincided with stop line; Described traffic lights surveyed area setting is used for the state of detection signal lamp, because the position of traffic lights is fixed, the interference of some images can be got rid of by the setting of traffic signal light condition surveyed area, to improve the accuracy detecting traffic signal light condition, the method for customization is the direct region matched by picture and traffic lights size on catadioptric mirror image parts of images taken by high-definition camera; Described track is customized for being set in direction multilane allowing vehicle operating, in some cities to clear stipulaties on multilane left lateral, craspedodrome, right lane, lane direction setting must be consistent with actual road junction roadway situation, the method of customization is that the road plane image directly taken by high-definition camera draws the lane line consistent with actual road junction roadway situation, and lane line needs to set up associating between lane line with virtual detection line after customizing; Described track associates with signal lamp and is customized for setting up associating between track with signal lamp, namely left lateral track associates with left lateral signal lamp, Through Lane associates with craspedodrome signal lamp, right lane associates with right lateral signal lamp, owing to establishing associating between lane line with virtual detection line above, therefore just establish associating of virtual detection line and signal lamp here; Just can judge whether vehicle also exists the behavior of making a dash across the red light and get over the act of violating regulations of line traveling according to such association and traffic law; In order to improve detection efficiency, the present invention is resolved into track running red light for vehicle detect road vehicle detection of making a dash across the red light, each track running red light for vehicle detects and uses a thread, namely road has how many tracks to correspond to how many threads and detects track respectively and to get on the bus a behavior of making a dash across the red light; By such customization by the image captured by the high-definition camera shown in accompanying drawing 3, be divided into the image on the left lateral track as shown in accompanying drawing 4, accompanying drawing 5 and accompanying drawing 6, the image on Through Lane and the image on right lane, in this three width image, include each autocorrelative virtual detection line and signal lamp;
In order to the convenience of the efficiency and calculating that improve detection, in software system design, in the present invention, the running red light for vehicle in each track is detected and calculate with a thread respectively, make use of the multiple threads technology in Java language, namely have how many tracks just to carry out the corresponding running red light for vehicle of computing with how many threads and detect; Accompanying drawing 7 is the running red light for vehicle detection procedure of some threads, as shown in the dotted portion in accompanying drawing 7, has how many threads so to run detection at the same time with regard to the dotted portion shown in drawings attached 7 arranged side by side;
Described microprocessor also comprises:
Color space conversion module, the image rgb color space for the signal lamp region to the traffic lights surveyed area setting described in the video image on each track is transformed into HSV color space, for pre-service is carried out in the state recognition of signal lamp; Colouring information is distinguished primarily of tone H, adds up tone H with color histogram, for distinguishing the color of signal lamp;
Signal lamp state detection module, for the detection to the signal lamp state in the video image on each track; First, by color normalized, 3 component values of HSV are all between 0 ~ l; The scope of the tone H of the green 3 kinds of color traffic lights of reddish yellow is as follows: red traffic light region: [0,0.0667] U [0.9058,1]; Amber signal lamp region: [0.0705,0.2353]; Green traffic lamp region: [0.2549,0.5882]; Then the number of pixels Number of statistical color histogram tone H in the green 3 kinds of color gamuts of reddish yellow is used
{ R, Y, G}, the total number of pixels of the pixel in traffic lights surveyed area is designated as Total, the ratio both calculating with formula (1),
NumRadio
{i}＝Number
{i}/Total i∈{R,Y,G} (1)
In formula, Number
{ i}for a certain color in the green 3 kinds of colors of reddish yellow is in the number of pixels of traffic lights surveyed area, Total is the total number of pixels of the pixel of traffic lights surveyed area, NumRadio
{ i}for the ratio of the total number of pixels of the pixel of a certain color in the number of pixels value and this region of traffic lights surveyed area in the green 3 kinds of colors of reddish yellow;
Finally carry out the condition adjudgement of traffic lights, determination methods is provided by formula (2),
In formula, NumRadio
{ R}for the ratio of the total number of pixels of the pixel of redness in the number of pixels value and this region of traffic lights surveyed area, NumRadio
{ Y}for the ratio of the total number of pixels of the pixel of yellow in the number of pixels value and this region of traffic lights surveyed area, NumRadio
{ G}for the ratio of the total number of pixels of the pixel of green in the number of pixels value and this region of traffic lights surveyed area;
Moving vehicle detection module, for the detection to the moving vehicle object in the video image on each track; Extract the foreground moving object on this track with motion history image algorithm, judge whether there is moving vehicle object on the virtual detection line on this track;
Motion history image algorithm, i.e. MHI algorithm, the motion history image obtained in MHI obtains after the adjacent image frame in a period of time interval carries out inter-frame difference and gray processing process, this algorithm obtaining motion history image can obtain motion outline template well for the target be kept in motion all the time and calculated amount is smaller, MHI algorithm can be used for creating a movement gradient image and the direction and the size that are calculated movement gradient by SOBEL operator, utilize this result can be further used for estimating the direction of motion of object simultaneously, just foreground moving object can be obtained because MHI algorithm does not need background modeling only to carry out inter-frame difference process, thus this algorithm has very high real-time for extraction foreground moving object, computing method are as shown in formula (3),
In formula, the binary picture sequence of D (x, y, t) moving region, τ is the duration, H
τthe motion history binary picture sequence that (x, y, t-1) was a upper moment;
Obtained the foreground moving object on track by MHI algorithm after, owing to track may also exist pedestrian or other Moving Objects, therefore need to judge whether these foreground moving object are vehicles; Foreground moving object is distinguished in the present invention by area threshold, shape and direction of motion, because the area of vehicle is larger than the area of pedestrian or cyclist, the direction of motion of pedestrian or cyclist is perpendicular to direction, track simultaneously, and shape has bigger difference, judge that computing method formula (4) represents
In formula, ∑ H
τthe sum of all pixels of the binary image that (x, y, t) is foreground object, T
1for area threshold, Shape is the shape attribute of foreground object, then finds shape attribute to obtain, T by searching border to binary image
2for rectangular shape threshold value, Direction is the direction of motion of foreground object, obtains, T by during calculating MHI algorithm
3for direction threshold value;
Effectively by the vehicle extraction that moves in track out, can then judge whether this moving vehicle is pressed in or crosses virtual detection line by the judgement of formula (4);
Whether vehicle peccancy judge module, having violating the regulations for detecting vehicle, substantially can be divided into two kinds of basic conditions here; Situation has the road clearly having lane traffic lamp, and another kind of situation is the road not having clear and definite lane traffic lamp and only have a traffic lights (reddish yellow is green).
The first situation is the road for there being clear stipulaties driveway travel directions, first vehicle lane change whether is had to travel according to virtual track is detected, if there is the vehicle on craspedodrome road to occur to turn right or turn left all to be judged to drive against traffic regulations, the situation of this situation and signal lamp has nothing to do, next is drives against traffic regulations relevant with signal lamp, this situation is will according to the situation of signal lamp, virtual track and stop the detection case in dummy line to make a decision, generally there is corresponding signal lamp for each track of each travel direction, and each travel direction also has corresponding stop line, therefore judge according to the traffic law that China is existing, be taken in certain track the when red of certain direction running time, if find that there is vehicle entering and stop dummy line or be in stopping just thinking to make a dash across the red light in dummy line position, as violating the regulations, when certain track be taken in the amber light of certain direction running bright time, entering stop dummy line as violating the regulations if find that there is vehicle,
The second situation is the road not having clear and definite lane traffic lamp and only have a traffic lights (reddish yellow is green), the judgement of this situation is more more complex than the first situation, even if this is because be allow to turn right according to the vehicle of the existing traffic law of China when red light on the right of road, that is, be not that all vehicles surmounting stopping dummy line position are all violating the regulations when red light, distinguish and whether have the result also will seeing vehicle tracking below violating the regulations; Specific practice finds that there is vehicle entering and stop dummy line or be in stopping that dummy line position just sets this vehicle and getting over line states, then continues to follow the tracks of, do not bend to right, so just as violating the regulations if follow the tracks of the vehicle finding to set more line states;
Accurately reflect that automobile driver makes a dash across the red light illegal activities process by the image information of more than three or three positions in the present invention, require that the information of first position clearly should be able to distinguish that motor vehicle is pressed in or crosses the situation of stop line, the image information of second and the 3rd position should be able to reflect that whole vehicle body has crossed the situation of stop line;
Judge to start when there is running red light for vehicle situation to record motor vehicle violation evidence obtaining video image in described vehicle peccancy judge module, start the image capturing first position; Along with passage of time, vehicle detected still towards under signal lamp disabled orientation ruuning situation, continue the image of candid photograph second and the 3rd position, the form capturing image is JPEG;
The license plate number identification module of vehicles peccancy, for identifying the license plate number of vehicles peccancy, automatically to carry out process violating the regulations; According to high-definition camera capture the vehicles peccancy image information of first position, the number on front part of vehicle car plate is identified; If the license plate number of vehicles peccancy can not be identified, continue to carry out license plate number identification by the vehicles peccancy image information of second position, when the vehicles peccancy image information of the 3rd position carry out license plate number recognition result still None-identified, license plate number is temporarily set to * * * * * * *, the manual confirmation of most relief break in traffic rules and regulations process department;
Record generation module violating the regulations, for by the whole process of vehicle peccancy behavior and same time, place, travel direction generate a record violating the regulations together with the information of the license plate number of vehicles peccancy, in record, include specific requirement and the Index Content of " the red-lamp running automatic recording system general technical specifications " defined specified in " the traffic safety illegal activities image forensics technical manual " and GA/T496-2004 international standard specified in GA/T832-2009 national standard; Capture first position, the image of second position and the 3rd position and video image all to record violating the regulations with attachment version apposition, annex name with time+mode in place names; Record violating the regulations and annex are kept in red light violation database of record;
Network transmission module, for the record violating the regulations that will be kept in red light violation database of record by the management system of Internet Transmission to law enforcement agency, to treat that the personnel of traffic police administrative authority carry out confirmation process.
Embodiment 2
With reference to Fig. 1 ~ Fig. 7, all the other are identical with embodiment 1, be not both and adopt filler lighting mode, when crossing illuminating ray deficiency, in order to capture the facial image on high-quality license plate image and driver's seat, the present invention is in candid photograph primary importance, the pulsed light adopting a kind of narrow-pulse generator to send during the image of the second place and the 3rd position carries out flash of light floor light, flash of light floor light has Microprocessor S3C44B0X, start flash of light floor light when candid photograph action simultaneously, pulsed light flash duration is 1/tens of general flash duration, flash of light floor light is arranged near monocular many detection faces video camera, flash illumination light projects towards stop line direction, crossing.Flash illumination light can be ignored the eye irritation of driver, but can photograph the car plate of vehicle and the face of driver clearly.
Claims (9)
1., based on a running red light for vehicle behavioral value device for the machine vision of the many detection faces of monocular, described running red light for vehicle behavioral value device comprises microprocessor and the monitoring sensor for monitoring crossing situation, and described microprocessor comprises:
View data reads and pretreatment module, passing the video image information of coming, carrying out pre-service to video image information for reading from monitoring sensor;
Virtual detection line setting, the setting of traffic lights surveyed area, track customization, track associate customized module with signal lamp, for customize track virtual stop line, customization traffic lights surveyed area, the track customized on road, customize track and associating between control signal lamp;
Network transmission module, for the image of the process violating the regulations by vehicles peccancy, when and where information by Internet Transmission to traffic police administrative authority;
It is characterized in that: described monitoring sensor is the high-definition camera of the many detection faces of monocular, the high-definition camera of the many detection faces of described monocular comprises the catadioptric minute surface in order to reflected signal lamp regional extent, in order to take the high-definition camera at stop line near zone travel condition of vehicle and signal lamp state, described catadioptric minute surface is positioned at the front upper place of high-definition camera, the corresponding signal lamp state in monitored crossing is photographed by the catadioptric high-definition camera function of catadioptric minute surface, the state near the stop line of monitoring crossing can also be photographed on the other hand by high-definition camera, high-definition camera imaging surface has the image information of two detection faces, and its middle and upper part is the image of signal lamp state, and bottom is the image of the crossing state corresponding with this signal lamp state, all tracks on the image pickup scope covering path of a high-definition camera, the fixing corner of adjustment catadioptric minute surface, makes the top in the image of high-definition camera imaging can photograph the image of signal lamp state,
Described microprocessor also comprises:
Color space conversion module, the image rgb color space for the signal lamp region to the traffic lights surveyed area setting described in the video image on each track is transformed into HSV color space;
Signal lamp state detection module, for the detection to the signal lamp state in the video image on each track;
Moving vehicle detection module, for the detection to the moving vehicle object in the video image on each track;
Whether vehicle peccancy judge module, also exist red light violation behavior for the vehicle detected on each track;
The license plate number identification module of vehicles peccancy, for identifying the license plate number of vehicles peccancy;
Record generation module violating the regulations, for by the whole process of vehicle peccancy behavior and same time, place, travel direction generate a record violating the regulations together with the information of the license plate number of vehicles peccancy;
The high-definition camera of the many detection faces of described monocular, high-definition camera and catadioptric minute surface is adopted to carry out integrated design, the supporting seat of high-definition camera draws a catadioptric mirror support bar, one end of support bar is connected with the universal auxiliary connection of catadioptric minute surface, and the lens location of the relative high-definition camera of catadioptric minute surface can adjust at any angle; When actual installation uses, by adjusting relative position and the angle of catadioptric minute surface and high-definition camera camera lens, make the top in the image of high-definition camera imaging can photograph the image of signal lamp state; Adjustment terminates the universal auxiliary connection of finger lock to ensure that high-definition camera function obtains the state of signal lamp effectively, accurately.
2. as claimed in claim 1 based on the running red light for vehicle behavioral value device of the machine vision of the many detection faces of monocular, it is characterized in that: described virtual detection line setting, the setting of traffic lights surveyed area, track customization, track associate in customized module with signal lamp, described virtual detection line setting is for setting up the virtual detection line corresponding with the stop line on real road, and the method for customization is that the road plane image directly taken by high-definition camera draws the line coincided with stop line; Described traffic lights surveyed area setting is used for the state of detection signal lamp, because the position of traffic lights is fixed, the interference of some images can be got rid of by the setting of traffic signal light condition surveyed area, to improve the accuracy detecting traffic signal light condition, the method for customization is the direct region matched by picture and traffic lights size on catadioptric mirror image parts of images taken by high-definition camera; Described track is customized for being set in direction multilane allowing vehicle operating, in some cities to clear stipulaties on multilane left lateral, craspedodrome, right lane, lane direction setting must be consistent with actual road junction roadway situation, the method of customization is that the road plane image directly taken by high-definition camera draws the lane line consistent with actual road junction roadway situation, and lane line needs to set up associating between lane line with virtual detection line after customizing; Described track associates with signal lamp and is customized for setting up associating between track with signal lamp, namely left lateral track associates with left lateral signal lamp, Through Lane associates with craspedodrome signal lamp, right lane associates with right lateral signal lamp, owing to establishing associating between lane line with virtual detection line above, therefore just establish associating of virtual detection line and signal lamp here; Just can judge whether vehicle also exists the behavior of making a dash across the red light and get over the act of violating regulations of line traveling according to such association and traffic law.
3. as claimed in claim 1 based on the running red light for vehicle behavioral value device of the machine vision of the many detection faces of monocular, it is characterized in that: in described signal lamp state detection module, first, by color normalized, 3 component values of HSV are all between 0 ~ l; Then the number of pixels Number of statistical color histogram tone H in the green 3 kinds of color gamuts of reddish yellow is used
{ R, Y, G}, the total number of pixels of the pixel in traffic lights surveyed area is designated as Total, the ratio both calculating with formula (1),
NumRadio
{i}＝Number
{i}/Total i∈{R,Y,G} (1)
In formula, Number
{ i}for a certain color in the green 3 kinds of colors of reddish yellow is in the number of pixels of traffic lights surveyed area, Total is the total number of pixels of the pixel of traffic lights surveyed area, NumRadio
{ i}for the ratio of the total number of pixels of the pixel of a certain color in the number of pixels value and this region of traffic lights surveyed area in the green 3 kinds of colors of reddish yellow;
Finally carry out the condition adjudgement of traffic lights, determination methods is provided by formula (2),
In formula, NumRadio
{ R}for the ratio of the total number of pixels of the pixel of redness in the number of pixels value and this region of traffic lights surveyed area, NumRadio
{ Y}for the ratio of the total number of pixels of the pixel of yellow in the number of pixels value and this region of traffic lights surveyed area, NumRadio
{ G}for the ratio of the total number of pixels of the pixel of green in the number of pixels value and this region of traffic lights surveyed area.
4. as claimed in claim 1 based on the running red light for vehicle behavioral value device of the machine vision of the many detection faces of monocular, it is characterized in that: in described moving vehicle detection module, extract the foreground moving object on track with motion history image algorithm, judge whether there is moving vehicle object on the virtual detection line on this track;
Described motion history image algorithm, i.e. MHI algorithm, the motion history image obtained in MHI obtains after the adjacent image frame in a period of time interval carries out inter-frame difference and gray processing process, this algorithm obtaining motion history image can obtain motion outline template well for the target be kept in motion all the time and calculated amount is smaller, MHI algorithm can be used for creating a movement gradient image and the direction and the size that are calculated movement gradient by SOBEL operator, utilize this result can be further used for estimating the direction of motion of object simultaneously, just foreground moving object can be obtained because MHI algorithm does not need background modeling only to carry out inter-frame difference process, thus this algorithm has very high real-time for extraction foreground moving object, computing method are as shown in formula (3),
In formula, the binary picture sequence of D (x, y, t) moving region, τ is the duration, H
τthe motion history binary picture sequence that (x, y, t-1) was a upper moment;
Obtained the foreground moving object on track by MHI algorithm after, owing to track may also exist pedestrian or other Moving Objects, therefore need to judge whether these foreground moving object are vehicles; Foreground moving object is distinguished by area threshold, shape and direction of motion, because the area of vehicle is larger than the area of pedestrian or cyclist, the direction of motion of pedestrian or cyclist is perpendicular to direction, track simultaneously, and shape has bigger difference, judge that computing method formula (4) represents
In formula, ∑ H
τthe sum of all pixels of the binary image that (x, y, t) is foreground object, T
1for area threshold, Shape is the shape attribute of foreground object, then finds shape attribute to obtain, T by searching border to binary image
2for rectangular shape threshold value, Direction is the direction of motion of foreground object, obtains, T by during calculating MHI algorithm
3for direction threshold value;
Effectively by the vehicle extraction that moves in track out, can then judge whether this moving vehicle is pressed in or crosses virtual detection line by the judgement of formula (4).
5., as claimed in claim 1 based on the running red light for vehicle behavioral value device of the machine vision of the many detection faces of monocular, it is characterized in that: in described vehicle peccancy judge module, be divided into two kinds of basic conditions; The first situation has the road clearly having lane traffic lamp, and the second situation is the road not having clear and definite lane traffic lamp and only have a traffic lights;
The first situation is the road for there being clear stipulaties driveway travel directions, first vehicle lane change whether is had to travel according to virtual track is detected, if there is the vehicle on craspedodrome road to occur to turn right or turn left all to be judged to drive against traffic regulations, the situation of this situation and signal lamp has nothing to do, next is drives against traffic regulations relevant with signal lamp, this situation is will according to the situation of signal lamp, virtual track and stop the detection case in dummy line to make a decision, generally there is corresponding signal lamp for each track of each travel direction, and each travel direction also has corresponding stop line, therefore judge according to the traffic law that China is existing, be taken in certain track the when red of certain direction running time, if find that there is vehicle entering and stop dummy line or be in stopping just thinking to make a dash across the red light in dummy line position, as violating the regulations, when certain track be taken in the amber light of certain direction running bright time, entering stop dummy line as act of violating regulations if find that there is vehicle,
The second situation is the road not having clear and definite lane traffic lamp and only have a traffic lights, the judgement of this situation is more more complex than the first situation, even if this is because be allow to turn right according to the vehicle of the existing traffic law of China when red light on the right of road, that is, be not that all vehicles surmounting stopping dummy line position are all violating the regulations when red light, distinguish and whether have the result also will seeing vehicle tracking below violating the regulations; Specific practice finds that there is vehicle entering and stop dummy line or be in stopping that dummy line position just sets this vehicle and getting over line states, then continue to follow the tracks of, if follow the tracks of the vehicle finding to set more line states not bend to right, so just as act of violating regulations.
6. as claimed in claim 1 based on the running red light for vehicle behavioral value device of the machine vision of the many detection faces of monocular, it is characterized in that: judge to start when there is running red light for vehicle situation to record motor vehicle violation evidence obtaining video image in described vehicle peccancy judge module, start the image capturing first position; Along with passage of time, vehicle detected still towards under signal lamp disabled orientation ruuning situation, continue the image of candid photograph second and the 3rd position, the form capturing image is JPEG.
7. as claimed in claim 1 based on the running red light for vehicle behavioral value device of the machine vision of the many detection faces of monocular, it is characterized in that: in the license plate number identification module of described vehicles peccancy, according to high-definition camera capture the vehicles peccancy image information of first position, the number on front part of vehicle car plate is identified; If the license plate number of vehicles peccancy can not be identified, continue to carry out license plate number identification by the vehicles peccancy image information of second position, when the vehicles peccancy image information of the 3rd position carry out license plate number recognition result still None-identified, allow the manual confirmation of break in traffic rules and regulations process department.
8. as claimed in claim 1 based on the running red light for vehicle behavioral value device of the machine vision of the many detection faces of monocular, it is characterized in that: in described record generation module violating the regulations, with the specific requirement and the Index Content that include " the red-lamp running automatic recording system general technical specifications " defined specified in " the traffic safety illegal activities image forensics technical manual " and GA/T496-2004 international standard specified in GA/T 832-2009 national standard in record; Capture first position, the image of second position and the 3rd position and the video image of process of making a dash across the red light all to record violating the regulations with attachment version apposition, annex name with time+mode in place names; Record violating the regulations and annex are kept in red light violation database of record.
9. as claimed in claim 7 based on the running red light for vehicle behavioral value device of the machine vision of the many detection faces of monocular, it is characterized in that: in candid photograph primary importance, the pulsed light adopting a kind of narrow-pulse generator to send during the red light violation behavior image of the second place and the 3rd position carries out flash of light floor light, flash of light floor light has Microprocessor S3C44B0X, start flash of light floor light when candid photograph action simultaneously, pulsed light flash duration is 1/tens of general flash duration, flash of light floor light is arranged near monocular many detection faces video camera, flash illumination light projects towards stop line direction, crossing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210551907.8A CN103065470B (en) | 2012-12-18 | 2012-12-18 | Detection device for behaviors of running red light of vehicle based on machine vision with single eye and multiple detection faces |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210551907.8A CN103065470B (en) | 2012-12-18 | 2012-12-18 | Detection device for behaviors of running red light of vehicle based on machine vision with single eye and multiple detection faces |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103065470A CN103065470A (en) | 2013-04-24 |
| CN103065470B true CN103065470B (en) | 2014-12-17 |
Family
ID=48108082
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210551907.8A Active CN103065470B (en) | 2012-12-18 | 2012-12-18 | Detection device for behaviors of running red light of vehicle based on machine vision with single eye and multiple detection faces |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103065470B (en) |
Families Citing this family (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103325258A (en) * | 2013-06-24 | 2013-09-25 | 武汉烽火众智数字技术有限责任公司 | Red light running detecting device and method based on video processing |
| CN104702907B (en) * | 2014-03-20 | 2019-05-10 | 杭州海康威视数字技术股份有限公司 | A kind of method and system of video-splicing |
| CN104134350A (en) * | 2014-08-11 | 2014-11-05 | 浙江力石科技股份有限公司 | Intelligent dome camera system for traffic violation behavior recognition |
| KR102383425B1 (en) * | 2014-12-01 | 2022-04-07 | 현대자동차주식회사 | Electronic apparatus, control method of electronic apparatus, computer program and computer readable recording medium |
| CN105812674B (en) * | 2014-12-29 | 2019-05-21 | 浙江大华技术股份有限公司 | A kind of color correct method, monitoring method and the device of signal lamp |
| JP6418089B2 (en) * | 2015-07-08 | 2018-11-07 | オムロン株式会社 | Image processing apparatus, traffic management system including the same, and image processing method |
| CN105070061B (en) * | 2015-08-19 | 2017-09-29 | 王恩琦 | Vehicle peccancy evidence obtaining checking method and its system |
| CN106560861A (en) * | 2015-09-30 | 2017-04-12 | 徐贵力 | Intelligent supervision method based on computer vision |
| CN107221160A (en) * | 2016-03-22 | 2017-09-29 | 深圳市以捷创新科技有限公司 | Smart electronicses police's system for detecting regulation violation |
| CN107221159A (en) * | 2016-03-22 | 2017-09-29 | 深圳市以捷创新科技有限公司 | The peccancy detection method of smart electronicses police's system for detecting regulation violation |
| CN107221164A (en) * | 2016-03-22 | 2017-09-29 | 深圳市以捷创新科技有限公司 | The peccancy detection method of smart electronicses police's system for detecting regulation violation |
| CN107230360A (en) * | 2016-03-24 | 2017-10-03 | 深圳市以捷创新科技有限公司 | The peccancy detection method of smart electronicses police's system for detecting regulation violation |
| CN107230359A (en) * | 2016-03-24 | 2017-10-03 | 深圳市以捷创新科技有限公司 | The peccancy detection method of smart electronicses police's system for detecting regulation violation |
| US9891628B2 (en) * | 2016-06-06 | 2018-02-13 | GM Global Technology Operations LLC | Sensor-based association of traffic control devices to traffic lanes for autonomous vehicle navigation |
| WO2018024320A1 (en) * | 2016-08-02 | 2018-02-08 | Bayerische Motoren Werke Aktiengesellschaft | A method for learning an illumination-pattern of a monitored automobile and a respective automobile |
| CN107786788A (en) * | 2016-08-31 | 2018-03-09 | 百度在线网络技术（北京）有限公司 | Intelligent traffic lamp system and its image processing method |
| CN107038869B (en) * | 2017-05-08 | 2020-01-21 | 钟辉 | Traffic operation violation distinguishing system |
| CN107067745A (en) * | 2017-06-15 | 2017-08-18 | 河南中安占海控股集团有限公司 | Intelligent traffic management systems |
| CN108280438B (en) * | 2018-02-02 | 2019-02-26 | 浙江东吴宏伟网络技术有限公司 | Entry region driver identity identifying system |
| CN110766949B (en) * | 2018-07-27 | 2021-09-28 | 浙江宇视科技有限公司 | Violation snapshot method and device |
| DE102018119024A1 (en) * | 2018-08-06 | 2020-02-06 | Knorr-Bremse Systeme für Nutzfahrzeuge GmbH | Camera surveillance system |
| CN111107305A (en) * | 2018-10-26 | 2020-05-05 | 杭州海康威视数字技术股份有限公司 | Monitoring interface configuration method, device and system and camera |
| US11157741B2 (en) | 2019-08-13 | 2021-10-26 | International Business Machines Corporation | Determining the state of infrastructure in a region of interest |
| CN110930715B (en) * | 2019-11-21 | 2021-07-20 | 浙江大华技术股份有限公司 | Method and system for identifying red light running of non-motor vehicle and violation processing platform |
| CN110929676A (en) * | 2019-12-04 | 2020-03-27 | 浙江工业大学 | Deep learning-based real-time detection method for illegal turning around |
| DE102019133171A1 (en) * | 2019-12-05 | 2021-06-10 | Saf-Holland Gmbh | System for checking an operating condition of a component on a vehicle and method for checking an operating condition |
| CN111325988A (en) * | 2020-03-10 | 2020-06-23 | 北京以萨技术股份有限公司 | Real-time red light running detection method, device and system based on video and storage medium |
| CN111968378A (en) * | 2020-07-07 | 2020-11-20 | 浙江大华技术股份有限公司 | Motor vehicle red light running snapshot method and device, computer equipment and storage medium |
| CN112200148B (en) * | 2020-11-06 | 2024-01-26 | 中国计量大学 | Decentralizing traffic integrity evaluation system |
| CN113870558A (en) * | 2021-09-26 | 2021-12-31 | 中国联合网络通信集团有限公司 | Vehicle detection method, vehicle detection device, vehicle, and traffic control device |
| CN115762173B (en) * | 2022-11-03 | 2023-08-22 | 湖北九州数字科技有限公司 | Traffic illegal behavior monitoring method and device based on three-dimensional road map |
| CN117392621B (en) * | 2023-11-07 | 2024-06-07 | 西南交通大学 | Method and system for identifying behavior of motor vehicle in case of turning right without giving away pedestrians |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6573929B1 (en) * | 1998-11-23 | 2003-06-03 | Nestor, Inc. | Traffic light violation prediction and recording system |
| CN100437660C (en) * | 2006-08-25 | 2008-11-26 | 浙江工业大学 | Device for monitoring vehicle breaking regulation based on all-position visual sensor |
| CN1987345B (en) * | 2006-12-20 | 2011-04-27 | 东华大学 | Single eye stereo vision sensor |
| CN101710448B (en) * | 2009-12-29 | 2011-08-10 | 浙江工业大学 | Road traffic state detecting device based on omnibearing computer vision |
| CN101930628A (en) * | 2010-09-21 | 2010-12-29 | 北京大学 | Monocular-camera and multiplane mirror catadioptric device-based motion capturing method |
| CN102110369B (en) * | 2010-12-21 | 2013-04-03 | 汉王科技股份有限公司 | Jaywalking snapshot method and device |
| CN102768800A (en) * | 2012-07-12 | 2012-11-07 | 复旦大学 | Motor vehicle red light running rule violation evidence obtaining method based on high-definition videos |
-
2012
- 2012-12-18 CN CN201210551907.8A patent/CN103065470B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN103065470A (en) | 2013-04-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103065470B (en) | Detection device for behaviors of running red light of vehicle based on machine vision with single eye and multiple detection faces | |
| CN110197589B (en) | Deep learning-based red light violation detection method | |
| CN103778786B (en) | A kind of break in traffic rules and regulations detection method based on remarkable vehicle part model | |
| CN102768801B (en) | Method for detecting motor vehicle green light follow-up traffic violation based on video | |
| CN103824452B (en) | A kind of peccancy parking detector based on panoramic vision of lightweight | |
| CN100437660C (en) | Device for monitoring vehicle breaking regulation based on all-position visual sensor | |
| AU2019235551B2 (en) | On-demand artificial intelligence and roadway stewardship system | |
| CN102903239B (en) | Method and system for detecting illegal left-and-right steering of vehicle at traffic intersection | |
| CN107154152A (en) | Video stake equipment and parking management method | |
| CN102496285A (en) | Method for determining red-light running of vehicles at intersection based on video detection and signal control system | |
| KR101742490B1 (en) | System for inspecting vehicle in violation by intervention and the method thereof | |
| US20180240336A1 (en) | Multi-stream based traffic enforcement for complex scenarios | |
| CN202422425U (en) | Video-detection-based intelligent signal control system for crossing | |
| CN101216996A (en) | A method for monitoring vehicle offences by video | |
| CN107240299A (en) | Autonomous land vehicle is to mobile object identification and the method for vehicle obstacle-avoidance | |
| CN102768800A (en) | Motor vehicle red light running rule violation evidence obtaining method based on high-definition videos | |
| CN102298844A (en) | Automatic rule breaking vehicle detection system and method | |
| CN104134350A (en) | Intelligent dome camera system for traffic violation behavior recognition | |
| CN105654734A (en) | Data processing method and device for vehicle violations | |
| JP6398920B2 (en) | Violator detection device and violator detection system provided with the same | |
| CN105046966A (en) | System and method for automatically detecting illegal parking behaviors in drop-off areas | |
| CN107447699A (en) | A kind of high-adaptability cycle path intelligent gate system | |
| CN107393311B (en) | A kind of license plate tamper Detection device and method | |
| KR102526583B1 (en) | Deep learning-based intersection object detection and automatic alarm system | |
| CN107123270A (en) | A kind of intelligent managing and control system in building site for slag-soil truck |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |