Summary of the invention
The present invention has designed a kind of freeway traffic event video detecting method, the defective that its " Indirect Detecting Method " and " direct detecting method " that has solved existing video traffic event detection exists separately,
In order to solve the technical matters of above-mentioned existence, the present invention has adopted following scheme:
A kind of freeway traffic event video detecting method may further comprise the steps:
Step 1: gather, detection system is gathered the highway video image by video capture device according to setting step-length, and video signal transmission is arrived the Video Events detecting device;
Step 2: the processing output road traffic accident of Video Events detecting device by three steps (comprise parkings, queuing, hypervelocity, retrograde etc.) and traffic parameter (comprising flow, car speed, vehicle classification etc.);
Step 3: event analysis is handled: the mode that detection system adopts vehicle tracking to detect and the detection of event algorithm combines is carried out event analysis and is handled, the road traffic accident that vehicle tracking detects by event detector output comes freeway incident is carried out analyzing and processing, the event algorithm detects by the road traffic parameter of event detector output freeway incident is carried out analyzing and processing, and both detect effect more accurately to reach at comprehensive contrast.
Further, three steps comprise video enhancing, video analysis and video understanding in the step 2; Wherein, video strengthens: video image is carried out noise estimate, and reduce in the video because the The noise that extraneous factor and human factor cause, for the video analysis layer provides high-quality video image, thereby improve Video processing result's reliability; Video analysis: detect by moving target, video analysis algorithms such as estimation and target following extract the space-time characteristics of objects in video bottom and middle level, for the event recognition in the high-rise Video processing provides the deduction foundation; Video is understood: finish identification to the monitor video event by the space-time object low-level image feature of analyzing and understand the video analysis layer and provide; Processing output road traffic accident by three steps (comprise parkings, queuing, exceed the speed limit and retrograde etc.) and traffic parameter (comprising flow, car speed and vehicle classification etc.).
Further, pass through in the step 3 to delimit the user's interest zone, and surveyed area is described (such as track, place traffic direction, whether turning etc.), analyze running status, the turning situation of vehicle; Utilize background subtraction, inter-frame difference and the frame interpolation moving object detection method of grading to detect the automobile storage of user's area-of-interest in situation; Utilize the auto model matching method to give vehicle classification; Utilize target object tracking, syntactic analysis and various forms of filtering and prediction estimation method to follow the tracks of vehicle, analyze the ruuning situation of estimation vehicle etc., utilize these information, can further analyze traffic density, queue length information.
Further, whether the change amount of the translational speed by detecting vehicle is higher than a certain threshold value, or judges whether its speed is lower than a certain particular value and judges whether it event has taken place.
Further, the calculating of video capture device setting angle in the step 1:
If the video camera setting height(from bottom) is H, its angle of view degree is α, and it is β that the elevation angle (video camera and vertical direction angle) is installed, and with the angle of road surface direction on vertical plane is
, α and β be surplus (β+α=90 °) mutually.With the angle of road surface direction on surface level be γ, width of roadway is W, S be ask angle, A is horizontal range; It is L that video monitoring requires detected link length, and the relation between these physical quantitys satisfies following formula:
Wherein: a=-htan α, b=-Ltan α, c=L-htan α;
Calculating the video camera elevation angle discusses in two kinds of situation:
(1) works as the angle of view
During more than or equal to A, video camera elevation angle β is:
2) work as the angle of view
During less than A, video camera elevation angle β is:
In a word, after calculating the video camera elevation angle, with the angle of road surface direction on vertical plane be:
For example: suppose field angle
=50 °
Through installing, debugging draws, and is 65 ° at video camera elevation angle β, when being 3.2 ° with angular separation, road surface γ, detects effective at the scene.
Further, the Video Events detecting device is to adopt motion detection, and image is handled, and realizes in target identification and the target following in working order that road traffic accident detects and the equipment of part DETECTION OF TRAFFIC PARAMETERS; The Video Events detecting device detects content for to handle and to analyze collecting the highway video image.
This freeway traffic event video detecting method has following beneficial effect:
The present invention handles and analyzes gathering the highway video image that comes in conjunction with the advantage of alert event and traffic parameter, judges highway communication event and warning in time by specific algorithm, can come the accurate recognition traffic events according to the image that collects.
Embodiment
Below in conjunction with Fig. 1 and Fig. 2, the present invention will be further described:
Fig. 1 at first detection system gathers the highway video image by video equipment according to setting step-length, and video signal transmission is arrived the Video Events detecting device; Secondly the processing output road traffic accident of Video Events detecting device by three steps (comprise parking, queuing, hypervelocity, retrograde etc.) and traffic parameter (comprising flow, car speed, vehicle classification etc.); For the type that makes the detection event is wider, precision is higher at last, and system adopts vehicle tracking to detect and the event algorithm detects the method that combines.The road traffic accident that vehicle tracking detects by event detector output comes freeway incident is carried out analyzing and processing, and the event algorithm detects by the road traffic parameter of event detector output freeway incident is carried out analyzing and processing.Both detect effect more accurately to reach at comprehensive contrast.
The algorithm that event automatically detects: by delimiting the user's interest zone, and surveyed area is described (such as track, place traffic direction, whether turning etc.), analyzes running status, the turning situation of vehicle; Utilize background subtraction, inter-frame difference and the frame interpolation moving object detection method of grading to detect the automobile storage of user's area-of-interest in situation; Utilize the auto model matching method to give vehicle classification; Utilize target object tracking, syntactic analysis and various forms of filtering and prediction estimation method to follow the tracks of vehicle, analyze the ruuning situation of estimation vehicle etc.Utilize these information, can further analyze information such as traffic density, queue length.
For the traffic monitoring video image, when vehicle generation event, such as cast anchor, during situation such as collision, its movement velocity will take place obviously to change or be reduced to zero rapidly.Therefore, whether the change amount of translational speed that can be by detecting vehicle is higher than a certain threshold value, or judges whether its speed is lower than a certain particular value and judges whether it event has taken place.
Fig. 2 Video Events detecting device is to adopt motion detection, and image is handled, and a series of technology such as target identification are middle in working order to realize that road traffic accident detects and the equipment of part DETECTION OF TRAFFIC PARAMETERS.Comprise video enhancing, video analysis and video understanding.
Video strengthens: video image is carried out noise estimate, and reduce in the video because the The noise that extraneous factor and human factor cause, for the video analysis layer provides high-quality video image, thereby improve Video processing result's reliability.
Video analysis: detect by moving target, video analysis algorithms such as estimation and target following extract the space-time characteristics of objects in video bottom and middle level, for the event recognition in the high-rise Video processing provides the deduction foundation.
Video is understood: finish identification to the monitor video event by the space-time object low-level image feature of analyzing and understand the video analysis layer and provide.
The calculating of video capture device setting angle:
If the video camera setting height(from bottom) is H, its angle of view degree is α, and it is β that the elevation angle (video camera and vertical direction angle) is installed, and with the angle of road surface direction on vertical plane is
, α and β be surplus (β+α=90 °) mutually.With the angle of road surface direction on surface level be γ, width of roadway is W, it is L that video monitoring requires detected link length, the relation between these physical quantitys satisfies following formula:
Wherein: a=-htan α, b=-Ltan α, c=L-htan α;
Calculating the video camera elevation angle discusses in two kinds of situation:
(1) works as the angle of view
During more than or equal to A, video camera elevation angle β is:
(2) work as the angle of view
During less than A, video camera elevation angle β is:
In a word, after calculating the video camera elevation angle, with the angle of road surface direction on vertical plane be:
For example: suppose field angle
=50 °
Through installing, debugging draws, and is 65 ° at video camera elevation angle β at the scene, and when being 3.2 ° with angular separation, road surface γ, it is relatively good to detect effect.
By reference to the accompanying drawings the present invention has been carried out exemplary description above; obvious realization of the present invention is not subjected to the restriction of aforesaid way; as long as the various improvement of having adopted method design of the present invention and technical scheme to carry out; or without improving design of the present invention and technical scheme are directly applied to other occasion, all in protection scope of the present invention.