Background technology
Traffic lights are the important composition compositions in the Traffic Systems.It is guaranteeing traffic safety, and standard vehicle behavior aspect has all played irreplaceable effect.Although set up the high speed loop in the Chinese large-sized city, removed a part of traffic lights from, in common street and crossroad, traffic lights are still one of Main Means of the various vehicle behaviors of standard.For the various traffic violations behaviors of stronger prevention, at the parting of the ways and some important location in street all kinds of cameras that cooperate supervision are often arranged, be mainly used in catching in violation of rules and regulations vehicle behavior, record license plate information etc.; Can say that aspect supervision vehicle behavior and license plate identification, computer vision has had significant progress.
Highway section pedestrian's street crossing control is the important content of urban traffic signal control, to how improving pedestrian crossing efficiency, at utmost reduces vehicle waiting delay time important in inhibiting simultaneously.Traditional highway section pedestrian's street crossing control method generally is to formulate the timing controlled scheme according to the historical pedestrian traffic amount situation in highway section or control experience, this kind control method changes under the comparatively stable situation comparatively applicable at the volume of traffic, the flow of the people of being expert at is difficult to adapt to transport need when changing greatly, can waste valuable traffic resource, affect traffic circulation efficient.
Application number is that 201010121624.0 Chinese patent discloses a kind of self-adapting type intelligent traffic light and control method thereof, it uses video camera to obtain the video image of all directions road, image is carried out information extraction, according to the density of waiting lines's vehicle and the density of waiting for the street pedestrian, traffic lights is carried out switching controls; Its control strategy is only applicable to vehicle flowrate and flow of the people roughly quite and under the more town road environment in crossroad; And for vehicle flowrate relatively less than under the less urban and suburban roads environment in flow of the people and crossroad, the traffic circulation efficient of this control mode is unsatisfactory, and computer vision is for vehicle and the dual identification of pedestrian and density analysis, calculated amount is large, and is reliably low.
Summary of the invention
For the existing above-mentioned technological deficiency of prior art, the invention provides a kind of highway section pedestrian's street crossing induction control method based on treating pedestrian's number, the traffic circulation efficient under can Effective Raise urban and suburban roads environment, and the control identification certainty is high.
A kind of highway section pedestrian's street crossing induction control method based on treating pedestrian's number comprises the steps:
(1) obtains the control threshold information;
(2) video image in row district is treated at two ends, Real-time Collection walkway, extracts to be in from video image to treat that the pedestrian that treats that row is distinguished counts information;
(3) according to controlling threshold information and treating that the pedestrian counts information, signal lamp is carried out switching controls.
Described control threshold information comprises: the longest green light displaying time of Vehicle Phase, the longest green light displaying time of pedestrian phases, Vehicle Phase to green light interval time between pedestrian phases, pedestrian phases to green light interval time, maximum between Vehicle Phase treats that pedestrian's number and minimum treat pedestrian's number.
In the described step (3), motor vehicle signal lamp and walker signal lamp are carried out the tactful as follows of switching controls:
A. working as Vehicle Phase is green light, when pedestrian phases is red light, judges whether Vehicle Phase green light displaying time reaches the Vehicle Phase maximum green perild:
If, to green light transition interval time between pedestrian phases, Vehicle Phase being switched to red light through Vehicle Phase, pedestrian phases switches to green light;
If not, and the walkway treat row district treat that pedestrian's number treats that greater than maximum the pedestrian counts the time, to green light transition interval time between pedestrian phases, Vehicle Phase is switched to red light through Vehicle Phase, pedestrian phases switches to green light;
B. working as Vehicle Phase is red light, when pedestrian phases is green light, judges whether the demonstration duration of walkway green light reaches the longest green light displaying time of pedestrian phases:
If, to green light transition interval time between Vehicle Phase, pedestrian phases is switched to red light through pedestrian phases, Vehicle Phase is switched to green light;
If not, and the walkway treats that the pedestrian's number for the treatment of in row district treats pedestrian's number less than minimum, to green light transition interval time between Vehicle Phase, pedestrian phases switched to red light through pedestrian phases, and Vehicle Phase switches to green light.
The present invention fully takes into account the actual traffic situation under the urban and suburban roads environment, take pedestrian crossing as top-priority object, obtain the information that pedestrian's number is treated on the walkway by Computer Vision Detection, and according to information signal lamp is carried out switching controls, can be when ensureing pedestrian's street crossing safety, reasonably excavate the time null resource of road traffic, effectively improve the traffic circulation efficient under the urban and suburban roads environment; Prior art is for vehicle and the dual identification of pedestrian relatively, and the present invention detects the recognition accuracy height, and the control reliability is strong.
Embodiment
In order more specifically to describe the present invention, below in conjunction with the drawings and the specific embodiments technical scheme of the present invention and control flow thereof are elaborated.
The object of present embodiment control is the signal lamp of car lane and crossing, walkway among Fig. 1, and this road environment is in urban and rural areas, and vehicle round on the car lane is less, and vehicle travels comparatively unobstructed in the track; And the both sides, track are respectively residential neighborhoods and free trade market, and have away from 1.5 kilometers from nearest crossroad, this crossing, on ordinary days in the pedestrian of both sides, track all travel to and fro between two places by this walkway.
For this road environment, present embodiment proposes a kind of highway section pedestrian's street crossing induction control method based on treating pedestrian's number, comprises the steps:
(1) obtains the control threshold information.
Obtain and comprise that the longest green light displaying time of Vehicle Phase, the longest green light displaying time of pedestrian phases, Vehicle Phase to green light interval time between pedestrian phases, pedestrian phases to green light interval time, maximum between Vehicle Phase treat pedestrian's number and the minimum control threshold value for the treatment of pedestrian's number, and according to these control threshold values, the control system of control Intersections is carried out Threshold.
(2) Real-time Collection video image and extraction treat that the pedestrian counts information.
Utilize two cameras respectively towards the row district for the treatment of at two ends, walkway, the video image in row district is treated at the Real-time Collection two ends, and video image is offered control system, is extracted from video image by control system to be in the pedestrian that treats who treats the row district and to count information.
Extraction treats that method that the pedestrian counts information mainly is by realizing for the detection of people's face and to the detection of people's shoulder regions; Concrete steps comprise:
1) obtains people's face positional information by the detection of people's face.The method that detects when forefathers' face is to use the Haar wave filter to detect.Its principle is at first to use the Weak Classifier input hundreds of width of cloth image of class Haar to train to one.This hundreds of width of cloth image comprises that about certain certain objects, for example people's face is called positive example; And hundreds of arbitrary image, be called counter-example.These training images need that all identical size is arranged.Here use the sorter that has trained that people's face is detected.The same method that adopts moving window travels through for all positions in the input picture, seeks human face region, obtains people's face positional information.
2) subtract each other by present image and background image and obtain the foreground pixel part.When the background image here refers to that nobody's face is in camera area, the background picture that this moment is corresponding.It is to keep constant in a certain time interval.Subtract each other by two two field pictures, obtain not identical pixel set.For this one part of pixel set, get pixel portion corresponding to human face region and gray processing, namely obtained foreground pixel.
3) extract edge feature.For the foreground pixel part, need to extract the feature that we are concerned about.What extract here is edge feature, and edge feature adopts the Canny operator to extract, and obtains a width of cloth bianry image, and it has represented the main profile information of people's face in the image.
4) mate with the shoulder template.Here select the shoulder feature of human body as the feature of judging the pedestrian, 2 reasons are arranged.Shoulder is in the higher position of health, and for the video camera that certain altitude is arranged, this is the information of the extreme higher position of the human body that can the most easily detect except face; Shoulder is with respect to other parts of health in addition, and the amplitude that arbitrarily swings is less, also can be used as one stable for detection of feature.When mating with the prospect profile image, the same moving window that changes size that adopts detects it, finally obtains pedestrian crowd's shoulder information.
5) for the people's face and the shoulder information that obtain, do inclusive-OR operation, to determine treating that the pedestrian in the row zone distributes and number.Here use the reason of inclusive-OR operation to be, for people's face and the shoulder information of same center, for it is a pedestrian; For other pedestrians of people's face or shoulder information that is blocked, as long as detect one of people's face or shoulder information, just think that there is the pedestrian in this place.This is in order to wait in the pedestrian is incorporated into all as far as possible.According to the final positional information of pedestrian, can obtain to treat the quantity information of pedestrian in the row zone.
(3) judge control switching signal lamp.
As shown in Figure 2, control system is according to the control threshold information and treat that the pedestrian counts information, by following control strategy motor vehicle signal lamp and walker signal lamp is carried out switching controls.
A. working as Vehicle Phase is green light, when pedestrian phases is red light, judges whether Vehicle Phase green light displaying time reaches the Vehicle Phase maximum green perild:
If, to green light transition interval time between pedestrian phases, Vehicle Phase being switched to red light through Vehicle Phase, pedestrian phases switches to green light;
If not, and the walkway treat row district treat that pedestrian's number treats that greater than maximum the pedestrian counts the time, to green light transition interval time between pedestrian phases, Vehicle Phase is switched to red light through Vehicle Phase, pedestrian phases switches to green light;
B. working as Vehicle Phase is red light, when pedestrian phases is green light, judges whether the demonstration duration of walkway green light reaches the longest green light displaying time of pedestrian phases:
If, to green light transition interval time between Vehicle Phase, pedestrian phases is switched to red light through pedestrian phases, Vehicle Phase is switched to green light;
If not, and the walkway treats that the pedestrian's number for the treatment of in row district treats pedestrian's number less than minimum, to green light transition interval time between Vehicle Phase, pedestrian phases switched to red light through pedestrian phases, and Vehicle Phase switches to green light.
In the present embodiment, the longest green light displaying time of Vehicle Phase is 90s, the longest green light displaying time of pedestrian phases is 60s, be 3s Vehicle Phase to green light interval time between pedestrian phases, be 10s pedestrian phases to green light interval time between Vehicle Phase, maximum treats that pedestrian's number is 12 people, and minimum treats that pedestrian's number is 4 people.