KR102558936B1 - System and method for controlling two-wheeled vehicles. - Google Patents

Abstract

The present invention relates to a two-wheeled vehicle enforcement system and method. More specifically, a two-wheeled vehicle photographing device installed on a horizontal member of a prop is provided so that a front camera having an inclined or wide-angle lens can continuously photograph the front of a two-wheeled vehicle entered by a front camera and the top of the two-wheeled vehicle when passing a reference line, and at the same time, the rear camera is provided to record the license plate of the rear of the two-wheeled vehicle, so that the analysis server tracks the two-wheeled vehicle based on the video taken through the front camera to change lanes of the two-wheeled vehicle It relates to a two-wheeled vehicle enforcement system and method capable of checking whether a passenger is riding with a still image extracted when passing a reference line while checking whether a helmet is worn or not on the head of a driver or passenger without misrecognition.

Description

{System and method for controlling two-wheeled vehicles.}

The present invention relates to a system and method for verifying whether a two-wheeled vehicle is complying with traffic rules or laws by installing a camera on a road. More specifically, it relates to a system and method for separately recognizing only a two-wheeled vehicle among a mobile vehicle and a two-wheeled vehicle traveling on a road, and then accurately checking whether a helmet is worn even if there is a passenger other than the driver riding the two-wheeled vehicle and recognizing the license plate number when not wearing a helmet.

In accordance with the current COVID-19 pandemic, the government is implementing social distancing policies to prevent the spread of the virus.

In other words, in accordance with the government's social distancing policy, we are minimizing the gathering of a certain number of people in one place.

As a result, many citizens order and eat various foods online through delivery agency services such as Coupang Eats and Baedal Minjok, rather than visiting department stores or marts in person.

Therefore, compared to before the outbreak of the previous Covid-19 (COVID-19) incident, the transaction volume through the delivery agency service has increased more than several times, and accordingly, the activities of two-wheeled vehicles (eg motorcycles) delivery drivers (riders) that deliver a lot of food have increased.

However, the number of two-wheeled vehicle (e.g., motorcycle) delivery drivers is limited, and in particular, many delivery drivers (riders) on the road ignore traffic signals / lanes and drive narrowly while delivering a large number of food in a limited time. The rate of traffic accidents is rapidly increasing.

For reference, a large amount of CCTVs or enforcement cameras are installed on the road to control vehicle movement (eg, 4-wheeled vehicles), but most of them recognize vehicle numbers from the front.

However, when the license plate is mounted on the front of the two-wheeled vehicle, there is a risk of accidents due to the steering wheel being twisted by wind, etc. Therefore, all two-wheeled vehicles are equipped with license plates only at the rear.

Therefore, even if a motorcycle driver does not wear a helmet and tries to crack down, in many cases unmanned enforcement is impossible because the number of the motorcycle cannot be recognized by CCTV or enforcement cameras installed on the road.

In particular, even if a speed camera is installed on the road to check the movement of the two-wheeled vehicle and whether or not the helmet is worn, misrecognition by the passenger often occurs when the driver and the passenger ride in close contact with the same motorcycle.

That is, when riding a two-wheeled vehicle, the driver in the front and the passenger in the back board in a state of close contact with each other side by side, and when the driving two-wheeled vehicle is photographed through a conventional CCTV or enforcement camera, an image in which the heads of the driver and the passenger overlap each other (eg, an image in which the head of the rear passenger is covered by the head of the front driver) is continuously filmed, resulting in misrecognition of whether the driver is wearing a helmet, or whether or not only the front driver is wearing a helmet, and whether or not the rear passenger is wearing a helmet cannot be confirmed. .

KR 10-2429312 B1 KR 10-185957 B1

Accordingly, an object to be achieved by the present invention is to check whether a driver riding a two-wheeled vehicle is wearing a helmet and to accurately recognize the vehicle number of the rear license plate of the two-wheeled vehicle when the helmet is not worn. In addition, another object of the present invention is to provide a system and method that can accurately determine whether or not a rider of a two-wheeled vehicle is wearing a helmet, as well as whether or not the passenger is wearing a helmet, which causes misrecognition.

Therefore, the present invention includes a horizontal member of a support, a two-wheeled vehicle filming device installed on the horizontal member of the support and capturing a video of a two-wheeled vehicle in motion, an analysis server for confirming the license plate number of the two-wheeled vehicle and the fact that the helmet is not worn based on the video filmed through the two-wheeled vehicle filming device, and a control server that monitors the video filmed through the two-wheeled vehicle photographing device and reports the vehicle number of the two-wheeled vehicle confirmed through the analysis server and the fact that the helmet is not worn,
The two-wheeled vehicle photographing device is provided with a front camera having an inclined or wide-angle lens to continuously photograph the front of the two-wheeled vehicle in motion and the upper part of the two-wheeled vehicle when passing through the reference line 15 formed in the direction directly below the front camera, and a rear camera provided to photograph the license plate by photographing the rear of the two-wheeled vehicle passing through the front camera,
The analysis server tracks the driving two-wheeled vehicle based on the video taken through the front camera to check whether the two-wheeled vehicle is changing lanes, and separates and recognizes only the head with the upper still image of the two-wheeled vehicle extracted when passing the reference line 15 formed directly below the front camera.

delete

delete

Therefore, the present invention can contribute to traffic order by separating and recognizing only the two-wheeled vehicle among mobile vehicles and two-wheeled vehicles traveling on the road, and then checking and regulating whether or not the driver of the two-wheeled vehicle is wearing a helmet.

In addition, the present invention not only solves the problem of inaccurately recognizing whether or not the driver is wearing a helmet by the passenger, but also checks if there is a passenger on the motorcycle and whether or not the passenger is wearing a helmet. It is also possible to check and enforce.

1 is a block diagram showing a system configuration according to an embodiment of the present invention;
2 is a block diagram of a two-wheeled vehicle photographing device according to an embodiment of the present invention.
3 is an exemplary view of a two-wheeled vehicle photographing device of a system according to an embodiment of the present invention.
4 to 6 are video screens analyzed by the analysis server.
7 is a search screen of a two-wheeled vehicle searched through the inquiry server of the present invention
8 is a flow chart showing a two-wheeled vehicle enforcement process of the present invention.

First, the terms used in the specification and claims of the present invention should not be construed as being limited to a dictionary meaning, and based on the principle that the inventor can appropriately define the concept of the term in order to best describe his/her invention. Based on the principle, it should be interpreted as meaning and concept consistent with the technical idea of the present invention.

Therefore, since the configurations according to the embodiments described in the specification and the configurations shown in the drawings are only preferred embodiments of the present invention and do not represent all the technical ideas of the present invention, it should be understood that they may be implemented in various ways by equivalents and modifications that can replace these configurations at the time of filing.

together "consists of" the present invention. or "formed." Etc. is not limited in a closed sense and "includes." Or it will have to be interpreted in an open sense, such as "to have".

Hereinafter, specific details for carrying out the present invention will be described with reference to the accompanying drawings.

1 briefly shows the configuration of a two-wheeled vehicle enforcement system according to an embodiment of the present invention, FIG. 2 briefly shows the configuration of a two-wheeled vehicle photographing device, FIG. 3 shows an example of a two-wheeled vehicle photographing device provided on a road, and FIGS.

Referring to this, the present invention is provided with a plurality of two-wheeled vehicle photographing devices 10 to photograph the two-wheeled vehicle 1 in motion, and the images captured through the two-wheeled vehicle photographing devices 10 are transmitted to a control center and monitored or monitored on-site.

At this time, the control center includes an analysis server 20 for analyzing the vehicle number of the two-wheeled vehicle 1 and confirming whether or not a helmet is worn based on the video captured by each two-wheeled vehicle photographing device 10, and monitoring the video taken through the two-wheeled vehicle photographing device 10, and a control server 30 that receives a report of the vehicle number of the corresponding two-wheeled vehicle along with the fact that the driver of the two-wheeled vehicle 1 confirmed through the analysis server 20 is not wearing a helmet, , The analysis server may be provided in the field according to an embodiment and transmit only the result value to the control server.

[0045] The two-wheeled vehicle photographing device 10 is a two-way photographing device equipped with a front camera 11 and a rear camera 13 so that the two-wheeled vehicle 1 or mobile vehicle in motion can be continuously photographed from the front and rear.

At this time, the two-wheeled vehicle filming device 10 is installed on the horizontal support member 5 at a height (e.g., 6 m) at which a large truck or a double-decker bus can safely pass in accordance with the Road Traffic Act, and the front camera is also inclined at a height that a large truck or a double-decker bus can safely pass through. It is to continuously photograph the upper part of the two-wheeled vehicle so that the driver or passenger's head can be confirmed.

In addition, the rear camera 13 of the two -wheeled car shooting device 10 is a camera having a resolution that can identify the license plate 3 of the two -wheeled vehicle 1 in which the two -wheeled vehicle 3 passes through the front camera 11 or the reference line 15, and the rear camera 13 treked through the front camera 11 The room is taken and the license plate of the two -wheeled vehicle is taken to check the video.

At this time, the two-wheeled vehicle photographing device 10 is equipped with an IR LED 12 that emits infrared rays in the 740 nm wavelength band so that the rear camera 13 can photograph one or more lanes and clearly photograph the license plate of the two-wheeled vehicle running even at night.

That is, in the rear camera 13, when light from the headlights of a moving vehicle following the two-wheeled vehicle is directly irradiated on the rear license plate of the two-wheeled vehicle, the light is reflected and applied to the rear camera 13, so that the license plate of the two-wheeled vehicle is clearly photographed by the IR LED 12 radiated obliquely from the left or right side with a visible light filter 13a provided inside to prevent the license plate of the two-wheeled vehicle from being photographed indistinctly.

In addition, the controller 14 can control the on / off operation of the front camera 11, the rear camera 13, and the IR LED 12, and the viewing direction or viewing angle of each camera 11 and 13. In particular, each video captured through each camera 11 and 13 is transmitted to a control server or analyzed on-site.

At this time, the controller 14 automatically operates the IR LED 12 in synchronization with the rear camera 13 according to the set time or illuminance value so that the license plate of the two-wheeled vehicle 1 can be clearly photographed.

In addition, the control center stores the images of the front camera 11 and the rear camera 13 transmitted from the two-wheeled vehicle filming device 10 into a database, and monitors or enforces road conditions through the analysis server 20 and the control server 30. It is a place. Representatively, it may be a CCTV integrated control center of each city or local government, or a 112 situation room or a 119 situation room, but is not limited thereto.

In addition, the analysis server 20, like the two-wheeled vehicle filming device 10, is provided at the site or in the control center, and in particular, tracks the two-wheeled vehicle 1 in motion based on video information captured through the front camera. In case the helmet 2 is not worn on the driver's head, it recognizes the vehicle number of the license plate photographed through the rear camera and reports it to the control server 30.

That is, as shown in FIG. 4, the analysis server 20 simultaneously photographs a plurality of lanes through the front camera 11, extracts an optimal still image 10a in which the two-wheeled vehicle 1 can be clearly seen from among videos of the moving vehicle 4 and the two-wheeled vehicle 1 moving along each lane, and then, from the extracted still images 10a, the object of the two-wheeled vehicle 1 among the two-wheeled vehicle 1 and the moving vehicle 4 through the two-wheeled vehicle dataset and the learner. After detecting only (20a), the running two-wheeled vehicle 1 is tracked.

At this time, the analysis server 20 tracks the two-wheeled vehicle 1 in motion so that the two-wheeled vehicle is not mistakenly recognized as a neighboring two-wheeled vehicle by changing lanes,

In particular, as shown in FIG. 5, when the two-wheeled vehicle 1 passes the reference line 15 formed on the road, a still image 10b is extracted, and the head object 20b of the driver or passenger is found and separated from the still image 10b, and then counted to determine whether the passenger is riding or not, and whether or not the helmet is worn.

At this time, the baseline 15 may be formed on the road, but may also be a virtual baseline created by the analysis server, and the learner of the analysis server 20 is a convolutional neural network (CNN) learner capable of deep learning based on image information, preferably an ecurrent neural network (RNN) learner or a YOLO learner,

When there is only one detected head object 20b, it is recognized that only the driver exists, and when there are two or more head objects 20b, it is recognized that one or more passengers exist.

And the analysis server 20 measures the degree of similarity through a deep learning learner based on the helmet dataset,

If the measured value is greater than or equal to the set value, it is determined that the helmet 2 or 2a is worn on the detected head object 20b of the driver or passenger, and conversely, if the measured value is less than or equal to the set value, it is determined that the helmet 2, 2a is not worn on the head object 20b of the driver or passenger.

In addition, as shown in FIG. 6, the analysis server 20 has the license plate 3 among the videos continuously photographed through the rear camera 13 when the motorcycle 1 passes the reference line or the front camera 11. After extracting the still image 10c in which viewing (viewing) is clear, the license plate 3 of the motorcycle 1 is found in the still image 10c and the license plate 3 of the motorcycle 1 is recognized.

At this time, the license plate (3) image is a single color image through which visible light is filtered, and is a vehicle number obtained by recognizing information extracted through a digit recognition algorithm when binarized based on boundary values.

In addition, the control server 30 displays the video taken by the front camera 11 and the rear camera 13 transmitted by the two-wheeled vehicle photographing device 10 so that the controller monitors the road conditions,

In particular, if a helmet is not worn on the head of the driver or passenger of the two-wheeled vehicle (1) driving on the road, the fact detected by the analysis server 20 (e.g., the fact that the helmet is not worn and the license plate number of the two-wheeled vehicle) is applied.

At this time, the analysis server 20 recognizes the speed of the two-wheeled vehicle using the time difference between the front camera and the rear camera continuously photographed based on the reference line formed on the road according to the embodiment, and can also control whether the two-wheeled vehicle is speeding.

In addition, the control server 30 or the analysis server 20 databases the still images 10a, 10b, 10c secured through the front camera 11 and the rear camera 13 based on the detected fact, and when the vehicle number is retrieved, the driver or driver and passenger may further include a query server 40 that can confirm / query the fact that the helmet is not worn on the head. At this time, the inquiry server may be a control server according to an embodiment.

That is, as shown in FIG. 7, the inquiry server 40 is based on the still images 10a, 10b, and 10c obtained through the two-wheeled vehicle photographing device 10, and when a search item such as a date and location is entered through a search window, the information of the two-wheeled vehicle photographed accordingly is loaded into the detail window and displayed. At this time, the detail window is loaded with the date, time, location, vehicle number, etc. of the two-wheeled vehicle driving on the road according to each item and displayed in a list. When one is selected, the images 10a, 10b, and 10c secured through the front and rear cameras are displayed when the two-wheeled vehicle passes, supporting the confirmation that the vehicle number of the two-wheeled vehicle 1 and the fact that the driver or passenger is not wearing a helmet are displayed.

8 illustrates a method for determining whether a passenger is riding in a two-wheeled vehicle and whether or not a helmet is worn by the passenger.

Referring to this description, in the present invention, the front camera and the rear camera of the two-wheeled vehicle photographing device 10 installed on the road separately or simultaneously photograph a moving vehicle or two-wheeled vehicle. (S100)

Preferably, the front camera 11 of the two-wheeled vehicle photographing device 10 simultaneously photographs a plurality of lanes obliquely or at a wide angle to capture an image of the front of the two-wheeled vehicle running and the top of the two-wheeled vehicle when passing the reference line, and the rear camera captures the front camera or the rear of the two-wheeled vehicle passing the reference line.

And the analysis server provided in the field or control center secures each image (front and rear) captured by the two-wheeled vehicle photographing device 10 (S110),

The analysis server creates a database that can be inquired according to a certain platform based on identification information (eg, IP of a two-wheeled vehicle photographing device), time, location, etc. (S120).

Among the images captured by the front camera of the two-wheeled vehicle photographing device 10, it is checked whether there is a two-wheeled vehicle running on the road, and if so, the corresponding two-wheeled vehicle is tracked (S130).

At this time, the analysis server extracts an optimal still image 10a whose outline is clearly visible according to filtering or image processing techniques from the video captured by the front camera 11, or makes the outline clear through a preprocessing process, and then detects the object 20a of the two-wheeled vehicle 1 among the two-wheeled vehicle 1 and vehicle 4 moving along each lane from the still image 10a through a two-wheeled vehicle dataset and a learner, and based on this, the object 20a of the two-wheeled vehicle 1 is traveling according to the change in time. The position of the two-wheeled vehicle 1 is tracked.

Then, the analysis server checks whether a passenger is riding in the corresponding two-wheeled vehicle based on the image taken at the top of the two-wheeled vehicle when the tracked two-wheeled vehicle passes the reference line (S140).

The reference line at this time may be a reference line formed on the road, but may also be a virtual reference line created by the analysis server. In particular, when the two-wheeled vehicle passes based on the reference line, a still image 10b of the upper end of the two-wheeled vehicle located directly below is extracted, and then the head object 20b of the driver or passenger is found and separated from the still image 10b, and the number is counted.

Then, the analysis server 20 checks whether a helmet is worn or not worn on one or two or more detected head objects 20b. (S150)

At this time, the analysis server 20 measures the degree of similarity through a deep learning learner based on the helmet data set. If the measured value is greater than or equal to a set value, it is determined that one or more of the detected head objects 20b have helmets 2 and 2a worn, respectively. Conversely, if the measured value is less than the set value, it is determined/recognized that the helmets 2 and 2a are not worn on the head object 20b of the driver or passenger.

And when the analysis server 20 recognizes that the helmet is not worn on the head object 20b, it checks the license plate of the two-wheeled vehicle and detects its vehicle number. (S160)

At this time, the analysis server 20 confirms the location of the two-wheeled vehicle tracked through the front camera, and extracts an optimal still image 10c in which the outline of the license plate 3 is clearly visible from the videos continuously photographed through the rear camera 13 when passing, or makes the outline visible through a preprocessing process, and then finds and separates the license plate 3 of the two-wheeled vehicle 1 from the still image 10c through an area value and shape recognition algorithm.

At this time, the still image 10c and the license plate 3 image are the light projected obliquely by the IR LED 12 while the headlights (reflected light) of the moving vehicle following are filtered by the visible light filter, and the vehicle number is clearly visible. It is a single color image, and the information extracted when it is binarized around the boundary value is detected through a digit recognition algorithm.

And the analysis server 20 reports the fact that the helmets 2 and 2a are not worn on the head of the driver or passenger riding the motorcycle and the license plate number of the motorcycle to the control server. (S170)

In addition, the control server 30 continuously monitors or supports unmanned enforcement based on the information reported through the analysis server.

So far, the present invention has been described with reference to the accompanying drawings, but this is only exemplary, and those skilled in the art can easily practice it by various modifications and equivalent theories. Therefore, the true technical protection scope of the present invention should be defined by the claims.

1: two-wheeled vehicle 2,2a: helmet
3: license plate 4: mobile vehicle
5: Horizontal material
10: two-wheeled vehicle filming device 11: front camera
12: IR LED 13: Rear Kemara
14: controller 15: reference line

Claims (5)

A two-wheeled vehicle filming device installed on the horizontal member of the support and a moving two-wheeled vehicle that is installed on the horizontal member of the support and photographs a moving two-wheeled vehicle, an analysis server that confirms the license plate number of the two-wheeled vehicle and the fact that the helmet is not worn based on the video filmed through the two-wheeled vehicle photographing device, and a control server that monitors the video filmed through the two-wheeled vehicle filming device and receives a report of the vehicle number of the two-wheeled vehicle confirmed through the analysis server and the fact that the helmet is not worn,
The two-wheeled vehicle photographing device is provided by a front camera having an inclined or wide-angle lens. It is provided to continuously photograph the upper part of the two-wheeled vehicle when passing through the front of the two-wheeled vehicle in motion and the reference line 15 formed in the direction directly below the front camera , while photographing the rear of the two-wheeled vehicle passing through the front camera to record the license plate.
The analysis server tracks the driving two-wheeled vehicle based on the video taken through the front camera to check whether the two-wheeled vehicle is changing lanes, and separates and recognizes only the head with the still image of the upper part of the two-wheeled vehicle extracted when passing the reference line 15 formed directly below the front camera . According to claim 1,
The analysis server detects an object of a two-wheeled vehicle among moving vehicles and two-wheeled vehicles in motion based on a video taken through a front camera, tracks its location, checks whether or not a lane is changed, and extracts a still image of the top of the two-wheeled vehicle when passing through the reference line. On the other hand,
The analysis server finds and separates the head object from the extracted still images, and counts the number to check whether a passenger is on board,
The analysis server measures the similarity of one or more head objects with the helmet data set stored through the learner, and recognizes that the driver or passenger's head is not wearing a helmet when the measured value is less than a set value. A two-wheeled vehicle enforcement system. According to claim 2,
When the analysis server recognizes that the helmet is not worn on the head of the driver or passenger, extracts a still image from the video taken through the rear camera, finds and separates the license plate object of the two-wheeled vehicle from the still image, and then recognizes the license plate number. A two-wheeled vehicle enforcement system. According to claim 1,
The analysis server recognizes the vehicle number with the image captured so that the vehicle number is clearly visible with the light projected obliquely by the IR LED in a state in which the headlights of the moving vehicle following behind are filtered by the visible light filter of the rear camera. Two-wheeled vehicle enforcement system. A step of separately or simultaneously photographing a moving vehicle or a two-wheeled vehicle by a front camera and a rear camera of a two-wheeled vehicle photographing device installed on the road;
The analysis server secures each image captured by the two-wheeled vehicle photographing device;
The analysis server comprises the steps of converting the obtained image into a database according to the designated platform;
The analysis server checks whether there is a two-wheeled vehicle driving on the road in the video taken through the front camera of the two-wheeled vehicle photographing device, and if there is, detects the two-wheeled vehicle and tracks its location, otherwise monitoring by the control server;
The analysis server extracts a still image of the upper part of the two-wheeled vehicle when it passes the reference line 15 formed directly below the front camera while checking whether the tracked two-wheeled vehicle has changed lanes, finds and separates the head object of the driver or passenger from the extracted still image, and counts the number of objects, and the analysis server determines that the passenger is on board if the number counted is two or more;
The analysis server measures the degree of similarity of one or two or more head objects with a learner deep-learning based on the helmet dataset, respectively, and if the measured value is greater than or equal to a set value, determining that the helmet is worn on the head object of the driver or passenger and, conversely, determining that the helmet is not worn on the head object of the driver or passenger if the measured value is less than or equal to the set value;
When the analysis server determines that the helmet is worn on the head object of the driver or passenger, it is monitored by the control server, and conversely, when it is determined that the helmet is not worn on the head object of the driver or passenger, the step of extracting a still image to confirm the license plate of the two-wheeled vehicle from among the video taken through the rear camera, and the step of finding the license plate of the two-wheeled vehicle from the extracted still image and recognizing the vehicle number;
The analysis server reports the vehicle number of the two-wheeled vehicle and the fact that the helmet is not worn to the control server.