KR101112562B1 - Image detecting apparatus and the method thereof - Google Patents

Abstract

An image detection system and a method thereof are disclosed. The present invention is to detect the vehicle information passing through the intersection, and in the intersection where there are many straight vehicles, but the left turn traffic is irregular, determines the number of vehicles in the left turn queue and the vehicle of the stop line, and transmits the determined information to the signal controller to control Therefore, the signal system can be operated safely and efficiently.

Description

Image detection system and its method {IMAGE DETECTING APPARATUS AND THE METHOD THEREOF}

The present invention is for detecting vehicle information passing through an intersection, and is an image detection system that can be used for signal control in place of a loop detector conventionally used as an intersection vehicle detector. The present invention relates to an image detection system and a method for detecting the number of vehicles in a left turn queue and the presence of a stop line at an intersection, and transmitting the determined information to a signal controller to efficiently control a signal.

In general, most of Korea's national roads have a naturally occurring form and composition, and thus have various characteristics for each region.The characteristics of these roads have a characteristic of crackdowns that are interrupted by traffic lights, but the distance between intersections is somewhat different. It is well known that it has a characteristic of continuous flow because of its long length.

In addition, most intersections of national roads are connected to small village roads, and these roads have very little traffic, but some national roads are in serious traffic problems due to the lack of proper traffic system due to large factories and new town development. Acquisition and proper signal control become very important.

To adjust traffic signal control patterns according to traffic conditions at intersections or driving roads so as to facilitate traffic traffic on roads, or to provide traffic users with road traffic information so that they can use roads that are easy to communicate. Information on vehicles driving on the road, that is, traffic information of the road is required.

For this purpose, a vehicle detector is installed on the road and traffic information such as traffic volume, driving speed, and occupancy of the road is obtained. As a vehicle detector for this purpose, a loop type detector and an image detector for detecting a vehicle by embedding a loop coil on the road Is being used.

Such a traffic signal control system using a loop type vehicle detector typically uses a loop coil embedded at an intersection to output an electrical signal indicating the presence or absence of a vehicle in a corresponding area, based on the output signal of the loop detector and the loop detector. It consists of a signal controller that detects the traffic conditions of the driving road) and outputs a corresponding control signal to the traffic light, and a plurality of roof coils are installed in a predetermined area around the intersection so that the vehicle can be located on the roof coil. When the on signal is output, the off signal is output. The loop controller processes the on / off signal input from the loop coil and outputs the signal to the signal controller side.

In addition, the loop detector counts the number of vehicles passing on the basis of the input on-signal, measures the holding time of the on-signal corresponding to the occupying time of the roof coil of the vehicle, and provides such information to provide the traffic congestion control, Calculate vehicle density and make it available for traffic signal control.

However, in the conventional traffic information detection technology, it is possible to determine the existence of the vehicle only when the vehicle passes correctly on the roof coil, and since the roof coil is a simple point detection sensor embedded in the road, the traffic condition or the vehicle may be used. There is a lack of flexibility in extracting information such as flow, speed, and congestion. In particular, it is difficult to intelligently respond to signals by grasping the number of vehicles waiting for the intersection.

In addition, the image detectors used in the past are mostly used for signal and speed regulation, and recently, they are also used for crime prevention, but they are used as image detectors for the intersection signal control which must be precisely controlled due to the vehicle detection error rate of the image detector. It is difficult.

In order to solve this problem, the present invention simultaneously uses an image algorithm for detecting the number of vehicles in a queue for a signal in a direction to be controlled by detecting vehicle information passing through an intersection and an image algorithm for determining whether a vehicle is in a stop line. The purpose of the present invention is to provide an image detection system and a detection method capable of minimizing the error rate of the image detector to ensure stability and to efficiently control the signal.

Another object of the present invention is to provide an image detection system and a method for detecting the same, which are capable of setting a detection area and outputting the number of vehicles in a queue and the presence or absence of a vehicle in a stop line detection area for each detection area in real time.

Another object of the present invention is to provide an image detection system for controlling intersection signals and a method of detecting the same, which can omit or prematurely terminate the signal manifestation.

The image detection system of the present invention for achieving the above object is an image photographing apparatus for photographing the image of the road, an image acquisition device for outputting the road image by converting the image photographed from the image photographing apparatus and the image acquisition device And a video detection device which sets a detection area from the road image output from the system and calculates and outputs in real time the number of vehicles in the queue and the presence / absence information of the vehicle in the stop line detection area using the set image detection algorithm for each detection area. It divides the image detection area into a queue detection area that can grasp the queue of vehicles and a stop line detection area that can determine the presence or absence of a vehicle stopped on a stop line. The error can be minimized.

The image detecting apparatus first sets an entire detection area including a left turn and a straight lane, a feature point registration area in a portion of a vehicle entering each lane, a left turn lane as a queue detection area, and an area of the queue detection area. A detection area setting unit for setting a stop line detection area in the exit portion and a detection information calculation unit for calculating information on the number of vehicles in the queue detection area set by the detection area setting unit and information on the presence or absence of vehicles in the stop line detection area; In order to calculate the number of vehicles in the queue detection area, the feature points of the vehicle are first extracted and registered, and the registered feature points are tracked and separated into individual vehicles.

In addition, in the image detection apparatus, the extraction of the vehicle feature point extracts and registers a vertex existing in the feature point registration area as a feature point of the vehicle, and the tracking of the feature point is performed by a predetermined size centering each of the feature points tracked up to the previous frame. The coordinates of the feature points tracked to the previous frame are moved to the point having the most similar value by comparing an area with a certain size centered on each feature point extracted from the current frame image, and the feature points are repeatedly tracked every frame. If there are no vertices present in the feature point registration area, points that are not separated into individual vehicles among the vertices registered and tracked before are regarded as the same vehicle and separated into individual vehicles, and vertices existing in the separated individual vehicles Their center of gravity as their center Considered to be an air to the center point of the vehicle of the separated vehicle is present in the queue detection area convoy vehicle and is characterized in that for calculating the number of vehicles in the queue.

In addition, the image detecting apparatus updates a background image in the stop line detecting region, generates a difference image, and then determines whether there is a vehicle in the stop line detecting region using the object size shown in the difference image, and the stop line The difference between the background image of the detection area and the current image of the stop line detection area is generated as a binary image, the size of the object shown in the binary image is calculated, and when the size is larger than a preset object threshold, the vehicle is placed in the stop line detection area. It is desirable to judge that there is.

A traffic signal controller for outputting a traffic light control signal to omit or prematurely terminate left turn display according to the number of vehicles in the queue and the presence or absence of the vehicle in the stop line detection area output from the image detection device, and a detection information transmitter for information transmission. It can be configured to include more.

On the other hand, the image detection method of the present invention for achieving this object comprises the steps of (a) photographing the image of the road with the image photographing apparatus, (b) setting the detection area from the road image, (c) the set Calculating in real time the number of vehicles in the queue and the presence or absence of vehicles in the stop line detection region by an image detection algorithm for each detection region; and (d) the number of vehicles in the queue and the stop line detection region calculated in step (c). The method may further include controlling a traffic light to omit or early terminate the left turn display according to the presence or absence information of the vehicle.

In the step (b), (b-1) setting the entire detection area including the left and right turning lanes, (b-2) setting the feature point registration area in the part where the vehicle enters in each lane, (b-3 Setting the entire left turn lane to the queue detection area; and (b-4) setting the exit portion of the queue detection area to the stop line detection area, wherein step (c) is performed. (c-1) first extracting and registering a feature point of the vehicle in order to calculate the number of vehicles in the queue detection area, (c-2) tracking the registered feature point, and (c-3) an individual vehicle Of course it can be made to include the step of separating and (c-4) calculating the number of vehicles in the queue.

In addition, in the step (c-1), the extraction of the vehicle feature point extracts a vertex existing in the feature point registration area as the feature point of the vehicle, and the tracking of the feature point in the step (c-2) is traced to the previous frame. The coordinates of the feature points tracked up to the previous frame are compared to the point having the most similar value by comparing an area of a certain size centered on each feature point and an area of the centered area of each feature point extracted from the current frame image. It moves and repeats it every frame to track feature points.

In the step (c-3), if there are no vertices present in the feature point registration area, points that are not separated into individual vehicles among the vertices registered and tracked before are regarded as the same vehicle and separated into individual vehicles. In the step (c-4), the center of gravity of the vertices of the separated individual vehicles is set as the center of the vehicle, and the center of the vehicle among the separated vehicles is present in the queue detection area. Considering the queued vehicle and calculating the number of vehicles in the queue, (c-5) generating the difference between the background image of the stop line detection region and the current image of the stop line detection region as a binary image, (c-6) After calculating the size of the object shown in the binary image, if the size is larger than the predetermined object threshold value, there is a vehicle in the stop line detection area It may be configured by further comprising stage.

Therefore, according to the image detection system and the method of the present invention, it is possible to calculate more accurate vehicle detection information that cannot be obtained by a point detector such as a loop detector, and by using two image detection algorithms, one image detection algorithm detects the detection. In case of failure, additional vehicle detection algorithm can ensure high vehicle detection reliability. This makes it possible to efficiently operate the signal control using the image detector.

In addition, according to the image detection system and method of the present invention, in order to compensate for the problem of the intersection that is inefficiently operated due to the existing signal control, it is possible to operate a safer and more efficient signal system by using the sensitive signal control. .

In addition, according to the image detection system and method of the present invention, when the traffic volume on the road is small or irregular, the traffic demand of the left turn is grasped and the response time is applied to minimize the loss time generated in the left turn signal of the road and the main road. It is possible to reduce the delay of intersection by smoothing traffic on the main road, and it is effective to prevent the intersection traffic accident caused by signal violation.

In addition, according to the image detection system and method of the present invention, since the number of vehicles in the queue and the presence or absence of the vehicle in the stop line detection region can be grasped by an algorithm set for each detection region, it is possible to control the traffic light more accurately. have.

1 is a main configuration of the image detection system of the present invention,
2 is a detailed configuration diagram illustrating an image detecting apparatus of the present invention;
3 is a detailed configuration diagram of a queue vehicle number information calculating unit of the present invention;
4 is a detailed configuration diagram of a vehicle presence information calculation unit in a stop line detection region of the present invention;
5 is a diagram showing a detection area setting of the present invention as an example;
6 is a view for explaining a process of separating the feature points tracked in the present invention into individual vehicles,
7 is a flowchart illustrating an image detection method of the present invention;
8 is a flowchart illustrating a vehicle detection method of a queue of the present invention;
And,
9 is a flowchart for explaining a vehicle detection method in the stop line region.

The terms or words used in this specification and claims are not to be construed as limiting in their usual or dictionary meanings, and the inventors may properly define the concept of terms in order to best explain their invention in the best way possible. It should be interpreted as meaning and concept corresponding to the technical idea of the present invention.

Throughout the specification, when a part is said to "include" a certain component, it means that it can further include other components, without excluding other components unless specifically stated otherwise. In addition, the terms “… unit”, “… unit”, “… device”, and “module” described in the specification mean a unit that processes at least one function or operation, which means hardware or software or hardware and software. It can be implemented as a combination of.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is a view illustrating the main configuration of the image detection system of the present invention, an image photographing apparatus 110 for photographing an image of a road, an image acquisition device 120 for outputting a digital road image, and an image acquisition device 120 It is possible to control the sensitive signal by setting the detection area from the road image output from the image detection device 200 and calculating the vehicle information for each detected detection area, and the detection signal calculated by the image detection device 200. It comprises a traffic signal controller 130 for outputting a traffic light control signal.

The image capturing apparatus 110 may be configured as a video camera to photograph and output the road image in the set detection area.

The detection area includes one or more image detection cameras and image detectors so that images of the entire road area including the left turn and the straight lane can be taken. Preferably, the road image may be photographed to determine whether the vehicle is stopped near the stop line of each left turn lane and to determine the number of vehicles waiting in the left turn lane area.

The detection area will be described in detail with reference to the drawings.

FIG. 5 is a diagram illustrating the detection area setting according to the present invention as an example. First, as shown in FIG. 5, the entire detection area 300 is set in a road area including a left and right turning lane, and a feature point is provided at a portion where a vehicle enters in each lane. In the registration areas 312 and 314, the entire left turn lane is set to the queue detection area 320, and the stop line detection area 330 is set to about the size of a vehicle at the exit of the queue detection area 320. After the detection area is set, the vehicle is detected by the set area whenever the image is detected.

These detection areas are determined by the image detection algorithm for each area in the image detection apparatus 200 to determine the number of queued vehicles or the presence or absence of vehicles in the stop line detection area.

In the present invention, the feature point registration areas 312 and 314 are divided into the feature point registration area 312 of the straight lane and the feature point registration area 314 of the left turn lane for convenience of description, but may be set as one feature point registration area.

The image acquisition apparatus 120 is configured to convert the road image photographed from the image photographing apparatus 110 into a digital image and output it to the image detection apparatus 200.

The image detecting apparatus 200 is configured to calculate the number of vehicles in the queue and the presence or absence of a stop line vehicle for each set detection region by a corresponding image detection algorithm. The present invention focuses on calculating and outputting the number of vehicles in the queue in the dead-rotation queue detection area 320 and the presence / absence information of the vehicle in the stop line detection area 330 in real time.

The detailed configuration of the image detection apparatus 200 will be described with reference to the drawings.

2 is a detailed configuration diagram illustrating an image detection apparatus of the present invention, Figure 3 is a detailed configuration diagram of the queue vehicle number information calculation unit of the present invention, Figure 4 is a vehicle presence information calculation in the stop line detection area of the present invention It is a detailed block diagram of the wealth.

As shown, the image detection apparatus 200 first sets the entire detection area 300 including the left and right turning lanes, and sets the feature point registration areas 312 and 314 in the area where the vehicle enters each lane in the entire detection area 300. And a detection area setting unit 210 and a detection area setting unit 210 for setting the entire left turn lane to the queue detection area 320 and the stop line detection area 330 at the exit of the queue detection area 320. The detection information calculating unit 220 for calculating the presence or absence of the vehicle information in the queue detection area 320 and the stop line detecting area 330 set in the above, and the detection area setting unit 210 and the detection information calculating unit 220. And a detection information transmitter 250 for providing information to the traffic signal controller 130 capable of controlling the response signal according to the number of vehicles in the queue and the presence or absence information of the vehicle in the stop line detection region.

At this time, the stop line detection area 330 is set to the size of about one vehicle at the exit of the queue detection area 320.

The detection information calculating unit 220 of the image detecting apparatus 200 includes a queue vehicle number information calculating unit 230 and a vehicle in the stop line detecting region 330 for calculating the number of queued vehicles in the queue detecting region 320. It is composed of a vehicle presence information calculation unit 240 in the stop line detection region for determining the existence.

 The queue vehicle number information calculating unit 230 first extracts and registers a feature point of the vehicle, and extracts and registers a feature point of the vehicle to separate and calculate the registered feature points into individual vehicles. 232, an individual vehicle separating unit 233, and a queue vehicle number calculating unit 234.

The vehicle feature point extraction and registration unit 231 extracts information of feature points that can be estimated as a vehicle and registers it as a tracking target, and registers a vertex existing in the feature point registration area 314 as a feature point of the vehicle and registers it as a tracking target. do. As a method of extracting vertices to be registered as a tracking target, various feature extraction image processing algorithms proposed to date can be used.

Referring to the drawing for explaining a process of separating the feature points tracked in the present invention of FIG. 6 into individual vehicles, first, the vehicle 3 in the drawings (a) and (b) is displayed as a vertex existing in the vehicle-"O". -These are extracted and registered as feature points as vertices-indicated by "★" in the feature point registration area 310 as they pass through the feature point registration area 310.

The feature point tracking unit 232 compares a region of a predetermined size of the current frame image and the previous frame image to track only the feature points existing in the feature point registration area 310 among the feature points extracted by the vehicle feature point extraction and registration unit 231. Use

In more detail, a region having a similar size is compared to a region having a certain size centering on each of the feature points tracked up to the previous frame and a region of a predetermined size centering on each of the feature points extracted from the current frame image. The coordinates of the tracked feature points are moved up to and repeated every frame to track the feature points.
In other words, the feature tracking is a comparison between a certain area centered on a feature point in a previous frame and a certain area centered on a feature point in a current frame, and thus, the most similar values, that is, the areas in the current frame with respect to the area of the previous frame. This means that the region having the smallest difference value is found and the feature point of the region found in the current frame is selected as the point created by moving the feature point of the previous frame. Here, the difference value is calculated by summing all differences of pixel values between corresponding pixels in the corresponding area. This is the same method used to search for motion vectors in video processing.

Referring to FIGS. 6C to 6D, a process of tracking vertices marked by “▲” that are not separated into individual vehicles while passing through the feature point registration area 310 is described.

The individual vehicle separating unit 233 is configured to separate the feature points tracked by the registered feature point tracking unit 232 for each vehicle, and if there are no vertices present in the feature point registration area, the individual vehicles among the vertices registered and tracked until before. The points that are not separated by the car are regarded as the same car and separated into individual cars.

6 (e) to 6 (e) show a process of separating the tracked feature points into individual vehicles.

 (e) If there are no vertices present in the feature point registration area 310 as shown in the drawing, among the vertices registered and tracked before, the points not separated into individual vehicles-(e) vertices indicated by "▲" in the drawing- It is regarded as the same vehicle and separated into a new vehicle, and the vehicle 2 and the vehicle 3 shown in (f) are separated into individual vehicles.

That is, (e) the drawing shows the case where no vertices exist in the feature point registration area in the state where vehicle 3 is not separated into individual vehicles, and (f) the vertices marked as “▲” that have been registered and tracked before in the drawing. Points which are not separated into individual vehicles are regarded as the same vehicle and registered separately as a new vehicle-a vertex marked by "

The separated individual vehicle sets the center of gravity as the center point of the vehicle so that the number of vehicles can be easily calculated in the next step.

The queue vehicle number calculating unit 234 calculates the number of vehicles as the center point of the vehicle. That is, the vehicle having the center point of the vehicle among the vehicles separated and tracked by the individual vehicle separator 233 is regarded as the queued vehicle and calculated as the number of vehicles in the queue.

The existence of the vehicle in the stop line detection area information calculating unit 240 updates the background image in the stop line detection area 330 and generates a difference image in order to determine the presence or absence of the vehicle in the stop line detection area 330. It is configured to determine the presence or absence of the vehicle in the stop line detection region 330 by using the object size shown in the image.

To this end, the presence / absence information calculation unit 240 of the vehicle in the stop line detection region includes a background image update unit 241 for updating a background image of the stop line detection region, a difference image generation unit 242 for generating a difference image, and an object. And a size calculation unit 243 and a presence / absence calculation unit 244 in the stop line detection area.

The background image updater 241 is configured to update the background of the stop line detection region 330. This is for calculating the difference between the current image and the background image after obtaining the background image of the stop line detection region 330.

The difference image generator 242 is configured to generate a difference between the background image updated by the background image updater 241 and the current image of the stop line detection region 330 as a binary image.

The binary image generated by the difference image generating unit 242 calculates the size of the object in the object size calculating unit 243, and then detects the stop line when the size of the binary image generating unit 244 is larger than the preset object threshold. It is determined that there is a vehicle in the area.

Object size calculation in the object size calculation unit 243 is calculated as the size of the object shown in the binary image, and presence or absence calculation unit 244 preferably sets the threshold in proportion to the vehicle size shown in the actual image.

In detail, the difference image is obtained from the background image and the current image, and the size of the object shown in the binary image is larger than the threshold value set by the user.

The threshold is set according to the size of the vehicle appearing in the actual image, but preferably set to 1/3 of the vehicle appearing in the actual image.

In other words, if the threshold value is increased, the vehicle similar to the color of the road may not be detected. If the threshold value is set too low, it may be determined that the vehicle exists even when there is no actual vehicle.

As described above, in the image detecting apparatus 200 according to the present invention, the number of vehicles existing in the queue detection area 320 is the number of vehicles having a center point in the queue detection area 320 among the vehicles separated by the feature point. Is calculated to calculate the number of queued vehicles, and it is determined that the vehicle exists in the stop line detection area 330 due to the difference between the background image and the current image in the stop line detection area 330.

In this way, the number of vehicles in the queue and the presence or absence of the vehicle in the stop line detection area calculated by the image detecting apparatus 200 are transmitted to the detection information transmitting unit 250 so as to omit the left turn display or early termination depending on the traffic volume when controlling the traffic light. To control.

For example, if there is no vehicle detected in the stop line detection area and the queue detection area on the main road, the left turn signal of the driveway may be omitted and the next signal may be displayed or detected in the stop line detection area and the queue detection area. If it is determined that the traffic of the vehicle is shorter than the preset left turn time, the left turn signal may be terminated early and the next signal may be displayed.

Hereinafter, an image detection method according to an embodiment of the present invention will be described with reference to the drawings.

7 is a flowchart illustrating an image detecting method of the present invention, FIG. 8 is a flowchart illustrating a vehicle detecting method of a queue of the present invention, and FIG. 9 is a flowchart illustrating a vehicle detecting method of a stop line region.

First, referring to FIG. 7, in the image detecting method of the present invention, an image of a road photographed by the image photographing apparatus 110 is received through a digital road image through the image obtaining apparatus 120 (S410).

When the digital road image is input, the image detecting apparatus 200 sets the road image as the entire detection area 300, but sets the entire left turn lane as the queue detection area 320, and the exit portion of the queue detection area 320. After setting the stop line detection area 330 to the size of one vehicle (S420), the number of queue cars in the queue detection area 320 is calculated (S430), and the vehicle presence information of the stop line detection area 330 is determined. After calculating (S440), the calculated detection information is transmitted to the traffic signal controller 130 (S450), and the left turn manifest is omitted according to the calculated number of vehicles in the queue and information on the presence or absence of vehicles in the stop line detection area. Controlling the traffic light to terminate early (S460).

That is, in step S420, the image detecting apparatus 200 sets the feature point registration area in the area where the vehicle enters in each lane, the entire left turn lane as the queue detection area, and sets the exit line of the queue detection area as the stop line detection area. It is.

In addition, the image detecting apparatus 200 first extracts and registers a vehicle feature point when a road image is input from the image obtaining apparatus 120 in order to calculate the number of queued vehicles in the queue detecting region 320 in step S430 (S431). ).

Extraction of the vehicle feature point extracts and registers a vertex existing in the feature point registration area 314 as the feature point of the vehicle.

The feature points extracted and registered in step S431 are followed by a tracking step to separate individual vehicles (S432).

The tracking of these feature points has the most similar value by comparing a certain amount of area centered on each of the feature points tracked up to the previous frame with a predetermined size area around each of the feature points extracted from the current frame image. By moving the coordinates of the feature points tracked up to the previous frame, the feature points are tracked by repeating it every frame.

After step S432, the feature points tracked by the feature point tracking unit 232 are separated by the individual vehicle separation unit 233 for each vehicle (S433).

When there is no vertex existing in the feature point registration area 314, such a vehicle is regarded as the same vehicle and separated into individual vehicles among the vertices registered and tracked before.

The image detecting apparatus 200 sets the center of gravity of the vertices existing in the individual vehicles separated in step S433 as the center of the vehicle, and the center of the vehicles among the separated vehicles exists in the queue detection area 320. The vehicle is regarded as a queued vehicle and the number of vehicles in the queue is calculated (S434).

Hereinafter, calculation of information on the presence or absence of a vehicle in the stop line detecting region will be described in detail with reference to the accompanying drawings.

In the stop line detection area, the information existence unit 240 detects the presence of the vehicle in the stop line detection area 330 first when a road image is input from the image acquisition device 120. In step S441, a difference image is generated by generating a difference between the updated background image and the current image of the stop line detection region 330 as a binary image (S442).

The presence / absence information calculating unit 240 of the vehicle in the stop line detecting region calculates the size of the object using the binary image generated in step S442 (S443).

When the size of the object is calculated in step S443, the size of the calculated object is compared with the preset object threshold (S444). If the size of the object is larger than the preset object threshold, the vehicle is located in the stop line detecting area 330. In operation S445, when the size of the object is smaller than the preset object threshold in operation S446, it is determined that there is no vehicle in the stop line detecting region 330.

After all, the present invention is to classify the road area by the detection area, and to determine the presence and number of vehicles in each area so that the traffic signal controller uses a variety of response signal control using an image detector.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art.

110: road image capture device 120: image acquisition device
130: traffic signal controller 200: image detection device
210: detection area setting unit 220: detection information calculation unit
230: waiting queue vehicle algebra information calculation unit
240: Vehicle presence information calculation unit in the stop line detection area
250: detection information transmitter

Claims (14)

delete delete delete delete An image photographing apparatus for photographing an image of a road;
An image acquisition device for converting an image taken by the image photographing device into a digital image and outputting a road image; and
An image detection device which sets a detection area from the road image output from the image acquisition device and calculates and outputs, in real time, the number of vehicles in the queue and the presence or absence information of the vehicle in the stop line detection area by means of an image detection algorithm for each detection area; Including;
The image detection device is
First, set the entire detection area including the left turn and the straight lane, and the feature point registration area in the area where the vehicle enters in each lane, the entire left turn lane as the queue detection area, and the stop line detection part in the exit of the queue detection area. A detection area setting unit for setting an area; and
A detection information calculation unit for calculating the presence / absence information of the vehicle in the queue detection area and the stop line detection area set by the detection area setting unit;
Including;
The image detection device is
In order to calculate the number of vehicle information in the queue detection area, the feature points of the vehicle are first extracted and registered, and the registered feature points are tracked and separated into individual vehicles, and calculated.
The extraction of the vehicle feature point extracts and registers a vertex existing in the feature point registration area as the feature point of the vehicle, and the tracking of the feature point is performed in a region of a predetermined size and the current frame image centered on each feature point tracked up to the previous frame. Compares the area with a certain size centered on each of the extracted feature points, finds the area with the smallest difference value, moves the coordinates of the feature points tracked to the previous frame to the feature point of the area found in the current frame, and repeats it every frame To track feature points,
If there are no vertices present in the feature point registration area, the points which have not been separated into individual vehicles among the registered vertices previously tracked are regarded as the same vehicle and separated into individual vehicles, and the weights of the vertices existing in the separated individual vehicles An image detection system comprising setting a center point as a center point of a corresponding vehicle and calculating a number of vehicles in the queue by considering a vehicle in which the center point of the separated vehicle exists in the queue detection area as a queue vehicle. .
delete 6. The method of claim 5,
The image detection device is
The difference between the background image of the stop line detection area and the current image of the stop line detection area is generated as a binary image, the size of the object shown in the binary image is calculated, and the stop line detection area when the size is larger than a preset object threshold. The image detection system, characterized in that it is determined that there is a vehicle.
6. The method of claim 5,
A detection information transmitter for providing information to a traffic signal controller capable of controlling a response signal according to the number of vehicles in the queue and information on the presence or absence of a vehicle in a stop line detection region;
Image detection system, characterized in that further comprises.

delete delete delete delete (a) taking an image of a road with an image photographing apparatus;
(b) setting a detection area from the road image;
(c) calculating in real time the number of vehicles in the queue and the presence or absence of vehicles in the stop line detection region using the set image detection algorithm for each detection region; and
(d) controlling the traffic light to omit or prematurely terminate the left turn display according to the number of vehicles in the queue and the presence or absence information of the vehicle in the stop line detection region calculated in step (c);
More,
Step (b) is
(b-1) setting an entire detection area including left and right turning lanes;
(b-2) setting a feature point registration area at a portion of the vehicle entering at each lane;
(b-3) setting the entire left turn lane as a queue detection area;
(b-4) setting an exit portion of the queue detection area as a stop line detection area;
Including;
Step (c) is
(c-1) first extracting and registering feature points of the vehicle in order to calculate the number of vehicles in the queue detection area;
(c-2) tracking the registered feature points;
(c-3) separating the individual vehicles; and
(c-4) calculating the number of vehicles in the queue;
Including;
Extracting the vehicle feature point in step (c-1)
The vertices existing in the feature point registration area are extracted as the feature points of the vehicle, and the tracking of the feature points in the step (c-2) is a region of a certain size centered on each of the feature points tracked up to the previous frame and the current frame image. Compares the area of the certain size centered on each of the feature points extracted from to find the area with the smallest difference value, and moves the coordinates of the feature points tracked up to the previous frame to the feature point of the area found in the current frame. Iteratively track feature points,
Step (c-3) is
If there are no vertices present in the feature point registration area, points which are not separated as individual vehicles among the vertices registered and tracked before are regarded as the same vehicle and separated into individual vehicles.
In the step (c-4), the center of gravity of the vertices of the separated individual vehicles is set as the center point of the vehicle, and the center of the vehicle among the separated vehicles is in the queue of the waiting matrix detection area. An image detection method, characterized in that it is regarded as a vehicle and the number of vehicles in the queue is calculated.
The method of claim 13,
Step (c) is
generating a binary image of the difference between the background image of the stop line detection region and the current image of the stop line detection region;
(c-6) calculating the size of the object shown in the binary image and determining that the vehicle exists in the stop line detection area when the size is larger than a preset object threshold;
Image detection method further comprises.

Images