CN103236191A - Video-based safety precaution method for vehicle merging from highway ramp - Google Patents

Abstract

The invention belongs to the field of video detecting, and provides a video-based safety precaution method for vehicle merging from a highway ramp. The method includes arranging two cameras in the same directions as two ways on highway entrances to detect moving vehicles, marking vehicle trajectories, acquiring actual moving distance through calculation according to the vehicle trajectories in a continuous tracking frame, acquiring actual speed of the vehicles, acquiring time difference of vehicle entering the merging areas between the two ways through the speeds of vehicles on the two ways, and judging possibility of vehicle collisions through the time difference. Compared with the prior art, the video-based safety precaution method is not limited by environments, capable of performing real-time and reliable precaution for all vehicles in the video areas, easy to realize, high in accuracy, applicable to real-time traffic precaution for highway ramps, and wide in application prospect.

Description

A kind of highway ring road vehicle based on video is incorporated safe early warning method into

Technical field

The invention belongs to the video detection range, be specifically related to a kind of highway ring road vehicle based on video and incorporate safe early warning method into.

Background technology

Recently, facing to the growth at full speed of China's highway mileage, traffic accidents increases year by year, and wherein highway ring road mouth is the Accident Area section.Particularly under the relatively poor situation of light, weather, traffic hazard very easily takes place in the ring road mouth.Highway ring road vehicle based on video is incorporated the rapid reaction capacity that the safe early warning technology has improved unusual condition into, simultaneously can with the data obtained in real time promptly and accurately be transferred to vehicle supervision department, increase work efficiency and the operation level of whole road network.

Existing method mainly is directly the transport information of ring road mouth to be judged by the driver, but the result is vulnerable to the subjectivity influence of human pilot, and not only reaction is slow, and influences such as light, weather are had stronger adaptability, can't satisfy the needs of practical application.

Summary of the invention

Defective and deficiency at prior art exists the objective of the invention is to, and provide a kind of highway ring road vehicle based on video to incorporate safe early warning method into, and this method can realize in real time vehicles all in the range of video, early warning reliably.

In order to realize above-mentioned task, the present invention adopts following technical scheme to be achieved:

A kind of highway ring road vehicle based on video is incorporated safe early warning method into, and this method is carried out according to following steps:

Step 1, the ring road mouth in the expressway sets up first video camera along the major trunk roads direction, and the range of first video camera is designated as L ₁Traffic video image on the first camera acquisition major trunk roads, be designated as first video image, adopt and based on the target vehicle tracking of feature in the video camera range region all to be driven towards all vehicles of incorporating the zone into and follow the tracks of the movement locus that obtains all cars;

Described target vehicle tracking based on feature is namely: adopt the frame-to-frame differences point-score based on block of pixels to detect moving vehicle, utilize the Moravec algorithm to extract corresponding angle point as the target location to detected moving image, and corresponding angle point is mated tracking;

Step 2 is set up second video camera at the ring road mouth first video camera place of setting up of expressway along the ring road direction, and the range of second video camera is designated as L ₂, the traffic video figure on the second camera acquisition ring road is designated as second video image, adopts based on the target vehicle tracking of feature in the video camera range region all to be driven towards all vehicles of incorporating the zone into and follow the tracks of the movement locus that obtains all cars;

Step 3, for first video image, calculate the actual travel speed of each vehicle, namely in any x two field picture, utilize the angle point of each car of Moravec algorithm, the interest value that compares each angle point, the point of screening interest value maximum, namely should the zone in the most tangible point of feature, as the position of each vehicle at the x two field picture; Adopt identical method, in the x+a two field picture, determine each vehicle in the position of x+a two field picture, the position of each angle point in the document image; Obtain the mapping relations between the capable and actual range of image pixel, namely mapping table obtains the physical location S of each vehicle in the x two field picture according to mapping table _i(i=1,2 ..., n), the physical location S in the x+a two field picture _i' (i=1,2 ..., n), then have:

$V_i=\frac{(S_i^{,}-S_i)V_0}{a}(i=\mathrm{1,2}...,n)$

In the formula:

V _iThe travel speed of-each vehicle, unit: frame/s;

V ₀-video playout speed, unit: frame/s;

The number of image frames of a-Continuous Tracking, unit: frame;

Step 4, for second video image, calculate the actual travel speed of each vehicle, namely in any y two field picture, utilize the method for step 3 to determine that each vehicle is in the position of y two field picture, determine each vehicle in the position of y+a two field picture at the y+a two field picture, the position of each angle point in the document image; Obtain the mapping relations between the capable and actual range of image pixel, namely mapping table obtains each vehicle physical location SS in the y two field picture according to mapping table _j(j=1,2 ..., m), the physical location SS in the y+a two field picture _j' (j=1,2 ..., m), then have:

${\mathrm{VV}}_j=\frac{({\mathrm{SS}}_j^{,}-{\mathrm{SS}}_j)V_0}{a}(j=\mathrm{1,2}...,m)$

In the formula:

VV _iThe travel speed of-each vehicle, unit: frame/s;

V ₀-video playout speed, unit: frame/s;

The number of image frames of a-Continuous Tracking, unit: frame;

Step 5, incorporate the actual range in zone and the actual travel speed of individual vehicle into according to each vehicle distance in first video image and second video image, obtain any vehicle of major trunk roads and sail into and incorporate on zone and the ring road arbitrarily that vehicle sails the mistiming of incorporating the zone into into into, be designated as t, then have:

$t=\frac{(L_1-S_i^{,})}{V_i}-\frac{(L_2-{\mathrm{SS}}_j^{,})}{{\mathrm{VV}}_j}(i=\mathrm{1,2}...,n,j=\mathrm{1,2}...,m)$

When | t|＞5s, then two vehicles enters safety when incorporating the district into;

When | t|≤5s, then two vehicles enters when incorporating the district into and may produce conflict, and shows the vehicle that may clash at the information table of major trunk roads and ring road.

Highway ring road vehicle based on video of the present invention is incorporated safe early warning method into, compared with prior art, is not subjected to environmental restraint, can to all vehicles in the range of video are carried out in real time, early warning reliably.And be easy to realize, accuracy is higher, is well suited for the real-time traffic early warning of highway ring road, have broad application prospects.

Description of drawings

Fig. 1 incorporates synoptic diagram into for highway ring road vehicle.

Fig. 2 is the synoptic diagram of the movement locus of each vehicle on the highway major trunk roads.

Fig. 3 is the synoptic diagram of the movement locus of each vehicle on the highway ring road.

Fig. 4 is highway major trunk roads vehicle and ring road vehicle collision detection process flow diagram.

Fig. 5 is highway ring road vehicle and trunk way vehicle collision detection process flow diagram.

Fig. 6 is the information table of highway major trunk roads, and this information table shows the vehicle that moves on the ring road.

Fig. 7 is the information table of highway ring road, and this information table shows the vehicle that moves on the major trunk roads.

Among the figure: black lines represents the highway lines, the vehicle that the little black surround representative of black may clash, and black is oval to represent the vehicle that moves on the highway, and the black line on the black ellipse represents the track of vehicle operating.

Below in conjunction with drawings and Examples content of the present invention is described in further detail.

Embodiment

Present embodiment provides a kind of highway ring road vehicle based on video and incorporates safe early warning method into, utilize the correlation technique that video detects and image is handled, obtain actual range and the actual travel speed of vehicle on highway major trunk roads and the ring road, its arrival is incorporated into the time in district and accurately calculated, thereby realization highway ring road vehicle is incorporated safe real-time early warning into.Need to prove and mapping table in the present embodiment adopt patent of invention " the video camera geometric calibration method under a kind of linear model " (open (bulletin) number: the video camera geometric calibration method CN102222332A) obtains.

To shown in Figure 7, the method for present embodiment specifically follows these steps to carry out as Fig. 1:

Step 1, the ring road mouth in the expressway sets up first video camera along the major trunk roads direction, and the range of first video camera is designated as L ₁Traffic video image on the first camera acquisition major trunk roads, be designated as first video image, adopt and based on the target vehicle tracking of feature in the video camera range region all to be driven towards all vehicles of incorporating the zone into and follow the tracks of the movement locus that obtains all cars;

Described target vehicle tracking based on feature is namely: adopt the frame-to-frame differences point-score based on block of pixels to detect moving vehicle, utilize the Moravec algorithm to extract corresponding angle point as the target location to detected moving image, and corresponding angle point is mated tracking;

Step 2 is set up second video camera at the ring road mouth first video camera place of setting up of expressway along the ring road direction, and the range of second video camera is designated as L ₂, the traffic video figure on the second camera acquisition ring road is designated as second video image, adopts based on the target vehicle tracking of feature in the video camera range region all to be driven towards all vehicles of incorporating the zone into and follow the tracks of the movement locus that obtains all cars;

Step 3, for first video image, calculate the actual travel speed of each vehicle, namely in any x two field picture, utilize the angle point of each car of Moravec algorithm, the interest value that compares each angle point, the point of screening interest value maximum, namely should the zone in the most tangible point of feature, as the position of each vehicle at the x two field picture; Adopt identical method, in the x+a two field picture, determine each vehicle in the position of x+a two field picture, the position of each angle point in the document image; Obtain the mapping relations between the capable and actual range of image pixel, namely mapping table obtains the physical location S of each vehicle in the x two field picture according to mapping table _i(i=1,2 ..., n), the physical location S in the x+a two field picture _i' (i=1,2 ..., n), then have:

$V_i=\frac{(S_i^{,}-S_i)V_0}{a}(i=\mathrm{1,2}...,n)$

In the formula:

V _iThe travel speed of-each vehicle, unit: frame/s;

V ₀-video playout speed, unit: frame/s;

The number of image frames of a-Continuous Tracking, unit: frame;

Step 4, for second video image, calculate the actual travel speed of each vehicle, namely in any y two field picture, utilize the method for step 3 to determine that each vehicle is in the position of y two field picture, determine each vehicle in the position of y+a two field picture at the y+a two field picture, the position of each angle point in the document image; Obtain the mapping relations between the capable and actual range of image pixel, namely mapping table obtains each vehicle physical location SS in the y two field picture according to mapping table _j(j=1,2 ..., m), the physical location SS in the y+a two field picture _j' (j=1,2 ..., m), then have:

${\mathrm{VV}}_j=\frac{({\mathrm{SS}}_j^{,}-{\mathrm{SS}}_j)V_0}{a}(j=\mathrm{1,2}...,m)$

In the formula:

VV _iThe travel speed of-each vehicle, unit: frame/s;

V ₀-video playout speed, unit: frame/s;

The number of image frames of a-Continuous Tracking, unit: frame;

Step 5, incorporate the actual range in zone and the actual travel speed of individual vehicle into according to each vehicle distance in first video image and second video image, obtain any vehicle of major trunk roads and sail into and incorporate on zone and the ring road arbitrarily that vehicle sails the mistiming of incorporating the zone into into into, be designated as t, then have:

$t=\frac{(L_1-S_i^{,})}{V_i}-\frac{(L_2-{\mathrm{SS}}_j^{,})}{{\mathrm{VV}}_j}(i=\mathrm{1,2}...,n,j=\mathrm{1,2}...,m)$

When | t|＞5s, then two vehicles enters safety when incorporating the district into;

When | t|≤5s, then two vehicles enters and incorporates the danger that Qu Shiyou produces conflict into, and shows the vehicle of the danger that produces conflict at the information table of major trunk roads and ring road.

Below provide specific embodiments of the invention, need to prove that the present invention is not limited to following specific embodiment, all equivalents of doing on present techniques scheme basis all fall into protection scope of the present invention.

Embodiment 1:

Ring road mouth in the expressway sets up first video camera along the major trunk roads direction, and the range of first video camera is L ₁=150 meters, traffic video image on the first camera acquisition major trunk roads, be designated as first video image, adopt and based on the target vehicle tracking of feature in the video camera range region all to be driven towards all vehicles of incorporating the zone into and follow the tracks of the movement locus that obtains all cars.The ring road mouth first video camera place of setting up in the expressway sets up second video camera along the ring road direction, and the range of second video camera is designated as L ₂=150 meters, traffic video figure on the second camera acquisition ring road, be designated as second video image, adopt and based on the target vehicle tracking of feature in the video camera range region all to be driven towards all vehicles of incorporating the zone into and follow the tracks of the movement locus that obtains all cars.Choose a vehicle arbitrarily for first video image, in the 10th frame, choose a angle point on this vehicle as its target location by screening, in the 30th frame, choose a angle point on this vehicle as its target location by screening, obtain mapping table according to a kind of labeling method, obtain S ₁=52 meters, S ₁The travel actual range S of 20 frames of this vehicle can be determined in '=100 meter ₁'-S ₁=48 meters, this vehicle is apart from the actual range L that incorporates the zone into ₁-S ₁'=50 meter.Choose a vehicle arbitrarily for second video image, in the 20th frame, determine the physical location of this vehicle by the screening angle point, in the 40th frame, determine the physical location of this vehicle by the screening angle point, obtain mapping table according to a kind of labeling method, obtain SS ₁=44 meters, SS ₁The travel actual range SS of 20 frames of this vehicle can be determined in '=60 meter ₁’ – SS ₁=16 meters, this vehicle is apart from the actual range L that incorporates the zone into ₂-SS ₁'=90 meter.So the number of image frames n=20 frame of Continuous Tracking, video playout speed V ₀=25 frames/s.Thereby obtain the speed V of major trunk roads vehicle ₁=60 meter per seconds, the speed VV of ring road vehicle ₁=20 meter per seconds, the major trunk roads vehicle sails into incorporates that vehicle sails the mistiming of incorporating the zone into on zone and the ring road into | t|=3.67 second＜5 second, think that vehicle enters on major trunk roads vehicle and the ring road and incorporate the danger that the district happens occasionally and conflicts into, information table at major trunk roads and ring road shows the vehicle that may clash, and human pilot is made judgement according to the real-time information of advices plate.

Embodiment 2:

Ring road mouth in the expressway sets up first video camera along the major trunk roads direction, and the range of first video camera is L ₁=150 meters, traffic video image on the first camera acquisition major trunk roads, be designated as first video image, adopt and based on the target vehicle tracking of feature in the video camera range region all to be driven towards all vehicles of incorporating the zone into and follow the tracks of the movement locus that obtains all cars.The ring road mouth first video camera place of setting up in the expressway sets up second video camera along the ring road direction, and the range of second video camera is designated as L ₂=150 meters, traffic video figure on the second camera acquisition ring road, be designated as second video image, adopt and based on the target vehicle tracking of feature in the video camera range region all to be driven towards all vehicles of incorporating the zone into and follow the tracks of the movement locus that obtains all cars.Choose a vehicle arbitrarily for first video image, in the 10th frame, choose a angle point on this vehicle as its target location by screening, in the 20th frame, choose a angle point on this vehicle as its target location by screening, obtain mapping table according to a kind of labeling method, obtain S ₁=90 meters, S ₁The travel actual range S of 20 frames of this vehicle can be determined in '=110 meter ₁'-S ₁=20 meters,, this vehicle is apart from the actual range L that incorporates the zone into ₁-S ₁'=40 meter.Choose a vehicle arbitrarily for second video image, in the 20th frame, determine the physical location of this vehicle by the screening angle point, in the 30th frame, determine the physical location of this vehicle by the screening angle point, obtain mapping table according to a kind of labeling method, obtain SS ₁=54 meters, SS ₁The travel actual range SS of 20 frames of this vehicle can be determined in '=60 meter ₁’ – SS ₁=6 meters, this vehicle is apart from the actual range L that incorporates the zone into ₂-SS ₁'=90 meter.So the number of image frames n=10 frame of Continuous Tracking, video playout speed V ₀=25 frames/s.Thereby obtain the speed V of major trunk roads vehicle ₁=50 meter per seconds, the speed VV of ring road vehicle ₁=15 meter per seconds, major trunk roads vehicle sail into incorporates that vehicle sails the mistiming of incorporating the zone into on zone and the ring road into | t|=5.2 second〉and 5 seconds, think that this vehicle of major trunk roads and this vehicle of ring road can not clash, two cars enters incorporates safety when distinguishing into.

Claims (1)

1. the highway ring road vehicle based on video is incorporated safe early warning method into, it is characterized in that this method is carried out according to following steps:

Step 1, the ring road mouth in the expressway sets up first video camera along the major trunk roads direction, and the range of first video camera is designated as L ₁Traffic video image on the first camera acquisition major trunk roads, be designated as first video image, adopt and based on the target vehicle tracking of feature in the video camera range region all to be driven towards all vehicles of incorporating the zone into and follow the tracks of the movement locus that obtains all cars;

Described target vehicle tracking based on feature is namely: adopt the frame-to-frame differences point-score based on block of pixels to detect moving vehicle, utilize the Moravec algorithm to extract corresponding angle point as the target location to detected moving image, and corresponding angle point is mated tracking;

Step 2 is set up second video camera at the ring road mouth first video camera place of setting up of expressway along the ring road direction, and the range of second video camera is designated as L ₂, the traffic video figure on the second camera acquisition ring road is designated as second video image, adopts based on the target vehicle tracking of feature in the video camera range region all to be driven towards all vehicles of incorporating the zone into and follow the tracks of the movement locus that obtains all cars;

Step 3, for first video image, calculate the actual travel speed of each vehicle, namely in any x two field picture, utilize the angle point of each car of Moravec algorithm, the interest value that compares each angle point, the point of screening interest value maximum, namely should the zone in the most tangible point of feature, as the position of each vehicle at the x two field picture; Adopt identical method, in the x+a two field picture, determine each vehicle in the position of x+a two field picture, the position of each angle point in the document image; Obtain the mapping relations between the capable and actual range of image pixel, namely mapping table obtains the physical location S of each vehicle in the x two field picture according to mapping table _i(i=1,2 ..., n), the physical location S in the x+a two field picture _i' (i=1,2 ..., n), then have:

$V_i=\frac{(S_i^{,}-S_i)V_0}{a}(i=\mathrm{1,2}...,n)$

In the formula:

V _iThe travel speed of-each vehicle, unit: frame/s;

V ₀-video playout speed, unit: frame/s;

The number of image frames of a-Continuous Tracking, unit: frame;

Step 4, for second video image, calculate the actual travel speed of each vehicle, namely in any y two field picture, utilize the method for step 3 to determine that each vehicle is in the position of y two field picture, determine each vehicle in the position of y+a two field picture at the y+a two field picture, the position of each angle point in the document image; Obtain the mapping relations between the capable and actual range of image pixel, namely mapping table obtains each vehicle physical location SS in the y two field picture according to mapping table _j(j=1,2 ..., m), the physical location SS in the y+a two field picture _j' (j=1,2 ..., m), then have:

${\mathrm{VV}}_j=\frac{({\mathrm{SS}}_j^{,}-{\mathrm{SS}}_j)V_0}{a}(j=\mathrm{1,2}...,m)$

In the formula:

VV _iThe travel speed of-each vehicle, unit: frame/s;

V ₀-video playout speed, unit: frame/s;

The number of image frames of a-Continuous Tracking, unit: frame;

Step 5, incorporate the actual range in zone and the actual travel speed of individual vehicle into according to each vehicle distance in first video image and second video image, obtain any vehicle of major trunk roads and sail into and incorporate on zone and the ring road arbitrarily that vehicle sails the mistiming of incorporating the zone into into into, be designated as t, then have:

$t=\frac{(L_1-S_i^{,})}{V_i}-\frac{(L_2-{\mathrm{SS}}_j^{,})}{{\mathrm{VV}}_j}(i=\mathrm{1,2}...,n,j=\mathrm{1,2}...,m)$

When | t|＞5s, then two vehicles enters safety when incorporating the district into;

When | t|≤5s, then two vehicles enters when incorporating the district into and may produce conflict, and shows the vehicle that may clash at the information table of major trunk roads and ring road.

Images