KR20010111202A - System for keep in order to wear seat belt and method thereof - Google Patents

Abstract

The present invention relates to a system for unfastening seat belts and a method thereof, and it is impossible to clamp seat belts with the naked eye on a highway or highway where the vehicle proceeds at a high speed because the conventional method for unfastening seat belts depends on the naked eye of a police officer. There was a problem that the driver's feeling of wearing the belt is lowered. In view of the above problems, the present invention includes an image capturing step of acquiring an image of a vehicle by detecting a front part of the vehicle by detecting that the vehicle has entered; A license plate area extraction step of extracting a license plate position of the vehicle from the captured image of the vehicle; A vehicle number detecting step of detecting a vehicle number from the extracted license plate area; A driver region extraction step of extracting an area where the driver is located by using the extracted license plate area or extracting an area where the driver is located by using the face of the driver from an entire image; A seatbelt determining step of detecting two diagonal lines parallel to each other in the extracted driver region and determining whether the diagonal lines are seat belts; If it is determined that the seat belt is determined in the seat belt determination step, the process is terminated, and if a non-seat belt or parallel diagonal line is not detected, the system continuously consists of a bill issuing step for issuing a bill for 24 hours a day. By checking whether the vehicle's seatbelt is worn, it can reduce the mortality rate and reduce the degree of injury when an accident occurs.

Description

Control system without seat belt and method thereof {SYSTEM FOR KEEP IN ORDER TO WEAR SEAT BELT AND METHOD THEREOF}

The present invention automatically checks whether the seat belt is worn by the driver, issues a penalty notice, and activates the seat belt of the driver, thereby reducing the accidents caused by a traffic accident. will be.

Currently in the United States, 50,000 people die each year and 3 million people are injured in traffic accidents, incurring significant social losses. Accident families are suffering from mental as well as economic damage. It has been widely known that wearing a seatbelt in a traffic accident can reduce the damage, and studies show that wearing a seatbelt can reduce the mortality rate by 57%. If all Europeans wear seat belts, 7500 lives per year can be saved in all of Europe.

In this regard, the US federal government has a stronger mandate to wear seat belts. In 1997, the United States issued a mandatory seat belt mandate that would allow drivers and occupants to wear seat belts, as well as allow police to detect drivers who did not wear seat belts. The mandatory announcement of the seatbelt includes new Hampshire, which currently does not require seatbelts, and most states where passengers who do not wear seatbelts do not want police to crack down. It includes sanctions to cut drastically. The United States is strongly urging to accept the 'primary seat belt laws' system, now in California and 11 other states, by September 30, 2002. In the road construction support, 1.5 percent are reported to be dedicated to the cost of the seat belt campaign. However, the state that the seat belt wearing rate of drivers and passengers exceeds 85% is excluded, and the US federal government seat belt wearing rate is only 68% nationwide.

In Korea, drivers and passengers are required to wear seat belts, but the crackdown method is determined only by the naked eye of police officers, and is not controlled by automated devices. In order to prevent speeding by automated devices, speeding control system is all for speeding prevention, which was introduced since 1990 and about 250 vehicles are installed and operated nationwide as of January 1, 2000. Speed violation control system using video is divided into local and central units. Local devices consist of cameras, lights, speed sensors, image processing devices, and transmission devices. The local unit and the central unit exchange data, video and system control signals through one-to-one dedicated lines. The central unit consists of a vehicle inquiry system, database server, output unit, communication unit and network unit. Speed Violation Enforcement Systems with Imaging Automatically perform a series of processes from speed measurement and imaging to billing. The whole process can be done automatically because the program of the image processing apparatus of the local apparatus automatically recognizes the license plate of the vehicle using a recognition algorithm and transmits the result to the central apparatus using the transmission apparatus. The central unit finds the address of the vehicle owner and the owner by using the vehicle number received, stores the result in a database, and issues a bill. The speed violation system using video is connected 24 hours online and all processes are automatic, so it has a very positive effect such as fairness of crackdown, reduction of police personnel and reduction of traffic accidents. We plan to install and run additional units.

As such, the device to control the speed violation is automated to induce the driver to be alert to the specified speed, but in the case of the non-belt wearing, there is no automated enforcement device, so police officers go directly to the road to crack down on the road. If the driver acknowledges the violation, he or she will sign the penalty notice and hand it over to the driver at the scene.

As described above, since the crackdown method without wearing a seatbelt depends on the naked eye of a police officer, it is impossible to clamp the seatbelt visually on a highway or highway where the vehicle proceeds at a high speed. Restraint of seatbelts is mostly restricted to city roads, but wearing seatbelts is more demanding on high-speed driving, and the way police are standing by the roadside and cracking down is not able to crack down on vehicles traveling in the inside lane. The evidence is insufficient due to the crackdown caused by the crackdown, and the driver's sense of duty to wear the seatbelt decreases due to the problem that the police officer and the driver are covering the fertilization ratio, or the driver who asks for a detention increases the traffic jam and impairs the fairness of the crackdown. there was.

In view of the above problems, the present invention provides a safety belt non-coupling enforcement system and a method for inspecting a driver's seat belt wearing regardless of a lane during high-speed driving, thereby increasing a driver's feeling of wearing a seat belt and reducing a death rate in an accident. Has its purpose.

1 is a configuration of the present invention seat belt non-regulation system.

2 is a detailed configuration diagram of the local apparatus in FIG. 1;

Figure 3 is a flow chart of the present invention seat belt non-coupling method.

4 is a flowchart of license plate recognition in FIG.

5A to 5C are flowcharts showing another embodiment of the driver area detection method in FIG.

Figures 6a to 6d is an embodiment of extracting the seat belt area through the corner emphasis method.

7 is a flow chart of seat belt recognition.

8A to 8C are photographs showing a process of extracting a driver's area from an image of an actual vehicle and extracting a seat belt area from the extracted driver's area;

* Description of the symbols for the main parts of the drawings *

10: Local device 11: Loop sensor

12: Loop card 13: Camera

14: trigger board 15: image processing device

16: DSU 17: dedicated line

30: central unit 31: integrated DSU

32: Router 33: Database Server

40: national institution network 50: vehicle query database server

The above object is to trigger the camera and the light when the vehicle is detected through the sensor and the trigger board to obtain an image of the detected vehicle, and extracting the license plate area of the vehicle from the acquired image to recognize the license plate and determine the position of the driver A plurality of local apparatuses including an image processing apparatus for extracting and determining whether the seat belt is worn, and a DSU for transmitting an output signal of the image processing apparatus through a dedicated line; Data transmitted from the plurality of local devices is authorized through a collective DSU and a router, and stored in a database server for storing the data and a vehicle inquiry database server connected to a national institution network using the data stored in the database server. After searching for the vehicle's track by a driver who is not wearing a seatbelt, the central unit is equipped with a vehicle tracking computer that outputs a bill through the output device, and detects that the vehicle has entered the front of the vehicle. An imaging step of acquiring an image of the vehicle by imaging the unit; A license plate area extraction step of extracting a license plate position of the vehicle from the captured image of the vehicle; A vehicle number detecting step of detecting a vehicle number from the extracted license plate area; A driver region extraction step of extracting an area where the driver is located by using the extracted license plate area or extracting an area where the driver is located by using the face of the driver from an entire image; A seatbelt determining step of detecting two diagonal lines parallel to each other in the extracted driver region and determining whether the diagonal lines are seat belts; If it is determined that the seat belt is determined in the seat belt determination step, the process is terminated, and if a parallel slant is not detected, the automatic seat control step is performed through a bill issuing step for issuing a bill. When described in detail with reference to the accompanying drawings, the present invention as follows.

1 is a block diagram of the present invention, without a seat belt control system, as shown in the loop card 11, when the vehicle is detected through a loop card (12) for outputting a signal that the vehicle is detected, and the loop card ( A trigger board 14 for acquiring an image of the vehicle detected when the camera 13 and the lighting (not shown) are operated according to the output signal of 12), and the license plate area of the vehicle is extracted from the acquired image. The image processing apparatus 15 which recognizes and extracts the position of the driver to determine whether the seat belt is worn, and the DSU which transmits the output signal of the image processing apparatus 15 to the central apparatus 30 through the dedicated line 17. A local device (10) comprising (16); Data transmitted from the plurality of local apparatuses 10 are received through the collective DSU 31 and the router 32, and are stored in the database server 33 and the database server 33 for storing the data. After searching the data stored in the vehicle inquiry database server 50 connected to the national institution network 40 to retrieve the vehicle's vehicle driven by a driver who is not wearing a seat belt, the report is output through the output device 35. Consists of a central unit 30, including a vehicle inquiry computer 34 for outputting the.

FIG. 2 is a detailed configuration diagram of the local apparatus 10 in FIG. 1, which includes a loop card 12 which outputs a control signal when a vehicle is detected from the loop sensor 11; A trigger board 13 for imaging the front side of the detected vehicle by operating the illumination 18 and the camera 14 which are powered by the relay board 19 according to the control signal of the roof card 12; ; A frame grabber 20 which detects a specific frame of the image of the vehicle captured by the camera 14 and generates an interrupt; A main memory 21 for storing the captured image of the vehicle; License plate area extraction means (22) for extracting the license plate area of the vehicle from the image data stored in the main memory (21); Character recognition means 23 for recognizing characters of the vehicle number from the extracted license plate area of the vehicle; Transfer the recognized vehicle number to the central unit 30 to find out the type of vehicle through vehicle inquiry, find out the distance between the driver and the license plate according to the type of vehicle, or extract the driver's face area Driver positioning means 24 for determining the position of the driver; Seat belt recognition means (25) for finding two lines located diagonally from the determined driver's position, and determining whether the seat belt is a seat belt by judging the width, the slope and the length of the line; The DSU 16 transmits the obtained image and data to the central device 30 through the dedicated line 17.

Hereinafter, the present invention will be described in more detail with the seat belt non-coupling system and the method configured as described above.

First, Figure 3 is a flow chart of the seat belt non-clamping method of the present invention, as shown in the loop sensor 11 when the vehicle enters, as shown in the loop card 12 generates a control signal. At this time, the operation principle of the loop sensor 11 flows a current through the loop cable to the loop coil and the lead wire, and when there is no vehicle, magnetic flux proportional to a constant inductance occurs. When a current flows in a conductor or wire, magnetic flux is generated around the conductor, and a magnetic field is formed and its direction follows Fleming's right hand rule.

In addition to the loop sensor, it is also possible to detect an entered vehicle by using an optical sensor, an image detector, a pressure sensor, a microwave, a laser, and an ultrasonic detector which are used for detecting a general vehicle.

When the vehicle passes over the magnetic field formed as described above, a change in magnetic flux occurs between the vehicle and the loop, and the entry of the vehicle is detected by using the change in the magnetic flux.

Next, the loop card 12 that detects the vehicle entering outputs a control signal, and the trigger board 13 receiving the vehicle is operated by the lighting 18 and the camera 14 to operate the front portion of the vehicle entering the vehicle. I will shoot.

At this time, since the illumination 18 is not necessary during the day, the relay board 19 is used to cut off the power applied to the illumination 18 during the day to prevent unnecessary power consumption.

Then, the image of the vehicle captured by the camera 14 is extracted by the frame grabber 20, the frame grabber 20 generates an interrupt to the main memory provided in the image processing device 15 The image data of the vehicle is stored at 21.

Then, the license plate area extraction means 22 extracts the area of the license plate from a video stored in the main memory 21 through a predetermined algorithm. At this time, the license plate extraction method is extracted using a point where the license plate area of the car has more contours than other parts.

This is obtained by using the line operation kernel described in Equation 1 below to obtain the contour image as a basic image for the license plate position extraction. The line operator kernel is a 1x3 gradient operator and is relatively noise-free.

[-1 0 1]

After horizontal gradient filtration, the image is binarized, and each object is extracted from the binarized image using the connection factor analysis. Each extracted style is compared with the size of the serial number of the license plate in advance to determine whether the object is a serial number character. If four serial number characters are found, the serial number position is calculated based on this, and the position of the remaining characters based on the serial number area of the vehicle is calculated and separated. Through this process, the starting position of the license plate and the size of the license plate are calculated to extract the license plate area.

Then, the character recognition means 23 separates each character from the extracted license plate area as shown in the character recognition flowchart shown in FIG. The characters at this time are divided into Korean letters and numbers as local letters, car model symbols, car use symbols and serial numbers.

In order to recognize the separated characters as described above, template matching is used. The matching method is a method of recognizing a number of letters and letters stored in advance.

Next, the driver positioning means 24 uses the table storing the license plate and the driver's separation distance for each vehicle type as shown in FIG. 5A, and calculates the driver's position by calculating the separation distance between the detected license plate and the driver. Will be extracted.

In this case, in order to know the type of the photographed vehicle, the detected vehicle number is output to the central unit 30 through the DSU 16 and the dedicated line 17, and the central unit 30 outputs the database of the vehicle. The vehicle model of the vehicle corresponding to the number is transmitted to the local apparatus 10 again.

Through this process, the type of the vehicle is recognized, and the distance between the license plate and the driver corresponding to the vehicle is found in the table to determine the position of the driver in the captured image.

In a method different from the method using the database, the driver's face region may be determined by extracting the driver's face region, which follows the order of FIGS. 5B and 5C.

That is, the face area of the driver is extracted by using the proposed face area detection system or algorithm, and the face area detection method in the black and white still image proposed or used up to now is a shape-based method and a feature-based method. , Based on the type, the shape-based method is to detect the face area using the contour of the face, the shape of the nose, the shape of the mouth, the shape of the eyes and eyebrows, etc. For example, this method is used to find a face by finding a curve of a face shape and a curve connecting eyebrows, eyes, nose and mouth, and a method of finding an ellipse according to a face region in a binarization image.

The feature-based method takes advantage of the fact that the geometrical relationship of the features of the eyes iris, the tip of the eye, the eye itself, the tip of the nose, the nostrils, the mouth, etc. is always constant on every face. When face recognition is performed using, distance between features is frequently used as a parameter of face recognition. This method is more flexible than other methods for changing facial expressions and tilting. In addition, many methods have been proposed to find features as a preprocessing process. Among them, a template matching algorithm using a template of a desired feature is widely used and known to be successfully used to extract a feature. There is a method of using a filter as a method that does not use a template, and in this case, the characteristics of variance in an image are often used, and a false recognition rate for a feature point may be larger than in other cases.

A representative example of the type-based method is a method of classifying images of face and non-face using neural networks. Over the years, many people have participated in this study, and neural networks with high recognition rates have been proposed. The reason why using neural networks has been successful is because they used a lot of training data. Apart from this, another method is to project the input image into a space that represents a significant change between already learned face images. An important feature in this method is the unique face.

When the face region of the driver is found in this manner, the driver region can be inferred from the face region.

Then, the seat belt recognition means 25 finds the seat belt in the set driver area. At this time, in order to find the seat belt in the lower part of the driver's face or the driver's area, the boundary area is found through an edge enhancement kernel and a line operation.

Equation 2 below is the sobel kernel, which is an edge-enhancing kernel, and the size of the found edge is calculated as Equation 3 and the direction is Equation 4 by the edge-enhancing kernel.

In this way, the image of the driver's area with the highlighted corner is binarized using the threshold. Threshold is one of the parametric methods and uses Otsu's algorithm, which is known to be relatively fast and computationally clear.

In order to find the seat belt in the binarized image, each component is separated using a connected component labeling algorithm. Connection element analysis is a method of erasing the objects that meet while reading the image horizontally and attaching each unique number in consideration of the connectivity in the new memory, which is shown in Figures 6a to 6d.

In other words, from the binary image shown in FIG. 6A, the connection element is analyzed to delete an object whose size is too small or too large compared to the seat belt, and the object is assigned a unique number. Obtain data on both sides of the belt.

Then, the result of FIG. 6B is thinned and converted into a line to obtain the result of FIG. 6C.

Next, the same number in FIG. 6c is connected by one line to extract the seat belt area of FIG. 6d.

Using the images of the two diagonal lines thus obtained, it is determined whether the diagonal line is the seat belt or not, and whether the separation distance between the two diagonal lines is the same as the width of the seat belt. It is determined whether it is distributed on the left and right with respect to the center, and it is determined whether the inclination of the two diagonal lines is in the range of about 40 to 50 degrees to determine whether it is a seat belt or not.

This process is shown in FIG. 7, when it is determined that the seat belt, the vehicle number and data obtained are not output, and the data is deleted.

FIG. 8 is a photograph illustrating a process of detecting a driver's area from an image of a vehicle photographed and binarizing the driver's area to detect whether a seatbelt is worn, and the boundary of the seat belt is clearly shown in the driver's area.

If two parallel oblique lines are not found in the driver's area, or when the oblique line is found but not the seat belt, the detected vehicle number and data are output to the central unit 30.

The central device 30 receives data sent from the local device 10 through the collective DSU 31 and the router 32, stores the data in the database server 33, and stores the data stored in the database server 33. After searching the data stored in the vehicle tracking database server 50 connected to the national institution network 40 through the vehicle tracking computer 34 to retrieve the vehicle tracks of the driver driving without a seat belt, The bill is output through the output device and charged to the driver who is not wearing a seat belt.

As described above, the present invention automatically checks whether the seat belt is worn on a vehicle which is operated at a position where the seat belt cannot be clamped by high speed driving such as a highway continuously for 24 hours a day, and automatically issues a bill. By inspiring the driver to wear the seat belt, thereby reducing the mortality rate in the event of an accident, there is an effect that can reduce the degree of injury.

Claims (6)

When the vehicle is detected through the sensor, a trigger board that acquires the image of the vehicle detected when the camera and the light are operated, extracts the license plate area of the vehicle from the acquired image, recognizes the license plate, and extracts the driver's position to wear the seat belt. A plurality of local devices including an image processing device for determining whether or not, and a DSU for transmitting an output signal of the image processing device through a dedicated line; Data transmitted from the plurality of local devices is authorized through a collective DSU and a router, and stored in a database server for storing the data and a vehicle inquiry database server connected to a national institution network using the data stored in the database server. After searching for the vehicle's vehicle by a driver who is not wearing a seat belt by searching for, the system is composed of a central unit having a vehicle tracking computer that outputs a bill through the output device. . The image processing apparatus of claim 1, further comprising: a main memory configured to store an image of a vehicle photographed through a camera; License plate area extraction means for extracting a license plate area of the vehicle from the image data stored in the main memory; Character recognition means for recognizing a character of a vehicle number from the extracted license plate area of the vehicle; Sends the recognized vehicle number to the central unit to find out the type of vehicle through track inquiry, finds the distance between the driver and the license plate according to the type of the vehicle, or extracts the driver's face area to determine the position of the driver. Driver positioning means for determining; Finding two lines diagonally located at the determined driver's position, and determining the width and inclination of the line and the length of the line to determine whether the seat belt is seat belt identification means characterized in that it is configured as a seat belt. system. An imaging step of detecting an entry of the vehicle and acquiring an image of the vehicle by imaging the front part of the vehicle; A license plate area extraction step of extracting a license plate position of the vehicle from the captured image of the vehicle; A vehicle number detecting step of detecting a vehicle number from the extracted license plate area; A driver region extraction step of extracting an area where the driver is located by using the extracted license plate area or extracting an area where the driver is located by using the face of the driver from an entire image; A seatbelt determining step of detecting two diagonal lines parallel to each other in the extracted driver region and determining whether the diagonal lines are seat belts; Ending when it is determined that the seat belt is in the seat belt determination step, if the non-belt or parallel oblique line is not detected, the seat belt is not attached, characterized in that the billing step issuing step for issuing a bill. The method of claim 3, wherein the step of extracting the driver's area checks the vehicle model using the detected vehicle number, and stores the license plate and the driver's separation distance for the vehicle model in a table storing the license plate and the driver's separation distance for each vehicle type. Safety belt non-clamping method, characterized in that for extracting the position of the driver using. The method of claim 3, wherein the step of extracting the driver's region is based on a method of detecting a face region by using a contour of a face, a shape of a nose, a shape of a mouth, a shape of an eye and an eyebrow, or a face of every person. Extracting the driver's position using a feature-based method that takes advantage of the fact that the geometric relationships of the iris, the tip of the eye, the eye itself, the tip of the nose, the nostrils, the mouth, etc. are always constant. The seat belt non-clamping method characterized by the above-mentioned. 4. The method of claim 3, wherein the determining of the seat belt comprises: detecting two parallel oblique lines using edge-enhanced line calculation in the extracted driver area; When the diagonal line is detected, it is determined whether the length of the two diagonal lines is appropriate to the length of the seat belt, and whether the separation distance between the diagonal lines is equal to the width of the seat belt, and whether both ends of the diagonal lines are located on the left and right sides from the center of the driver's area, respectively. Safety belts do not wear, characterized in that the step consisting of determining whether the belt worn.