TWI824265B - Detection system and violation detection method for vehicles - Google Patents

Abstract

The present invention proposes a detection system for vehicles, which includes an electronic device and a server connected to the electronic device. The electronic device includes a detection module, a processing module and a transmission module; the server includes a status judgment module, an event processing module and a database. Among them, the state determination module includes a state determination unit that determines the motion state of the vehicle; a signal recognition unit that recognizes whether at least one image information includes a traffic signal; a lane recognition unit that recognizes whether the vehicle enters the auxiliary lane; and determines whether the vehicle enters the auxiliary lane. Violation detection unit to detect whether a vehicle is driving illegally or violating intersection clearance rules. In addition, a violation detection method for vehicles is also disclosed.

Description

Detection system and violation detection method for vehicles

The invention relates to a vehicle-mounted detection system, and in particular, to a vehicle-mounted detection system capable of signal identification, image processing and area determination.

With the rapid development of intelligent transportation, logistics and transportation management will become more efficient. However, general logistics management systems cannot automatically detect vehicle violations during vehicle dispatch tasks, such as illegal left turns, illegal vehicle turns, or vehicles staying at intersections. It is difficult for managers to assess the daily work performance of on-site drivers. Under evaluation, there is still room for improvement.

In order to solve the problems mentioned in the prior art, the present invention proposes a detection system for vehicles, which includes an electronic device and a server connected to the electronic device. The electronic device includes a detection module, a processing module and a transmission module; the server includes a status judgment module, an event processing module and a database. Among them, the state determination module includes a state determination unit that determines the motion state of the vehicle; a signal recognition unit that recognizes whether at least one image information includes a traffic signal; a lane recognition unit that recognizes whether the vehicle enters the auxiliary lane; and determines whether the vehicle enters the auxiliary lane. Violation detection unit to detect whether a vehicle is driving illegally or violating intersection clearance rules.

Furthermore, the detection module includes an image detection unit, a status detection unit and a positioning unit; the processing module includes a computing unit and a module control unit; and the transmission module includes an information Transceiver unit.

Furthermore, the module control unit includes a detection module control unit and a transmission module control unit. The detection module control unit regularly stores the image information, the motion status information and the positioning information; the transmission module control unit dynamically controls the transmission rate of the transmission module according to the speed of the vehicle.

Furthermore, the information transceiver unit includes an image transceiver unit and a text transceiver unit. The image transceiver unit dynamically controls the transmission rate of the image information according to the speed of the vehicle; the text transceiver unit transmits the motion status information and the positioning information.

Furthermore, the electronic device further includes an application program, which includes a user interface, a vehicle real-time monitoring interface, an event query interface, a vehicle historical trajectory interface, and a system setting interface.

The present invention also proposes a violation detection method for vehicles, which includes: real-time detection of at least one image information of the vehicle's driving path; detection of an acceleration angle of the vehicle, and a positioning unit detecting positioning information of the vehicle; Determine whether the vehicle has turned based on the acceleration angle and the positioning information; obtain all image information before a specific time, determine whether all the image information contains illegal signs; and determine whether the vehicle violates the rules.

The present invention further proposes another violation detection method for vehicles, which includes: determining whether the vehicle is located in an intersection area based on a geographical map system; determining whether the vehicle is in an idling state based on the speed of the vehicle; and real-time detection of the driving path of the vehicle. at least one image information, and based on the at least one image information Determine whether the vehicle has entered the auxiliary lane based on the signal; determine whether the vehicle violates the regulations based on whether the vehicle is located in the intersection area, whether the vehicle is idling and whether the vehicle has entered the auxiliary lane.

The above brief description of the present invention is intended to provide a basic explanation of several aspects and technical features of the present invention. The Summary of the Invention is not a detailed description of the invention, and therefore its purpose is not to specifically enumerate key or important elements of the invention, nor to define the scope of the invention. It is merely to present several concepts of the invention in a concise manner.

1:Detection system

10: Electronic devices

12:Detection module

122:Image detection unit

124: Status detection unit

126: Positioning unit

14: Processing module

142:Computing unit

144: Detection module control unit

146:Transmission module control unit

16:Transmission module

162: Information sending and receiving unit

164: Communication control unit

20:Server

22: Status judgment module

222: Status judgment unit

224: Number identification unit

226: Lane recognition unit

228: Violation detection unit

24:Event processing module

26:Database

The first figure is a schematic diagram of a detection system according to a preferred embodiment of the present invention.

The second picture is a schematic diagram of the camera device.

The third figure is a pre-processing flow chart for processing the artificial intelligence model input image in this embodiment.

The fourth figure is a flow chart of the lane identification method according to the preferred embodiment of the invention.

The fifth figure is a flow chart of violation detection according to the preferred embodiment of the present invention.

The sixth figure is an algorithm flow chart for intersection clearance violation detection according to the preferred embodiment of the present invention.

The seventh figure is a flow chart of the violation detection method according to the first embodiment of the present invention.

Figure 8 is a flow chart of a violation detection method according to the second embodiment of the present invention.

In order to understand the technical features and practical effects of the present invention and implement it according to the contents of the description, the preferred embodiment as shown in the drawings is further described in detail as follows:

The present invention proposes a detection system for vehicles, which obtains vehicle detection information through image, status and positioning detection units in electronic devices, and transmits the detection information through a transmission module. Transmitted to an external server or cloud service for signal identification, image processing and area determination; thereby, when a defined violation occurs, the backend system will notify the user of the electronic device, solving the problem of vehicle dispatch in the existing fleet management system During the mission, traffic violations such as illegal left turns, illegal turns, and vehicles staying at intersections cannot be detected.

First, please refer to the first figure, which is a schematic diagram of a detection system according to a preferred embodiment of the present invention. As shown in the first figure, the vehicle detection system 1 of this embodiment includes an electronic device 10 and a server 20 connected to the electronic device 10 . Specifically, the electronic device 10 is, but is not limited to, a smart phone, a mobile device, a tablet computer, a personal computer or a wearable device and other electronic devices with networking, display and operation functions. The server (or cloud server) can be remotely connected to the electronic device through the base station using 4G or 5G mobile communication technology, thereby receiving the vehicle information detected by the electronic device 10 in real time and sending the processed violation notification.

The detection system 1 for vehicles in this embodiment will be further described below. First, the electronic device 10 further includes a detection module 12, a processing module 14 and a transmission module 16. The detection module 12 includes an image detection unit 122, a status detection unit 124 and a positioning unit 126. Among them, the image detection unit 122 can be, but is not limited to, a camera device or an infrared sensor, which can real-time capture image information during the driving process of the vehicle as an image input source for signal recognition and assisted turning lane recognition. The state detection unit 124 may be, but is not limited to, a gravity sensor (G-Sensor, also known as an accelerometer), a gyroscope or a magnetometer, or may be a nine-axis attitude sensor or an inertial measurement unit (Inertial Measurement Unit, IMU); among them, the nine-axis attitude sensor has a three-axis gyroscope, a three-axis magnetometer and a three-axis accelerometer, and the inertial measurement unit is composed of three accelerometers and three gyroscopes. The purpose of the state detection unit 124 is to detect the angle, position, displacement, rotation and speed of the vehicle. Status information is used as a basis for vehicle turning and rotation. The positioning unit 126 obtains the positioning information of the vehicle through GPS as a basis for vehicle positioning, vehicle speed calculation and intersection violation area determination.

The processing module 14 is connected to the detection module 12 and includes a module control unit and a calculation unit 142 . Among them, the calculation unit 142 can calculate the speed of the vehicle based on the positioning information. The module control unit further includes a detection module control unit 144 and a transmission module control unit 146; the detection module control unit 144 is The regularly executed program regularly stores the image information, motion status information and positioning information obtained by the detection module, and the transmission module control unit 146 dynamically controls the transmission rate of the transmission module according to the speed of the vehicle. In this embodiment, the processing module includes a microcontroller unit (MCU), or any device that can accept the transfer of information between modules and perform calculation processing and then send control commands, falls within the scope of this protection. Inventions should not be limited to this.

The function of the processing module 14 in the system is to obtain detection information, calculate the vehicle speed and control the transmission module. Specifically, the calculation unit 142 can calculate the current speed of the vehicle through positioning information, date and time; the vehicle speed is calculated using the longitude, latitude and time obtained by the electronic device, and the method is based on two time points before and after (assumed to be t The time difference between ₂ and t ₁ ) is calculated by the longitude and latitude distance between the two time points (based on Pythagorean theorem: a ² + b ² takes the square root). The calculation formula is as follows:

The detection module control unit 144 accesses the detection module through regularly executed programs to obtain information including images, motion status, positioning, date and time, etc. It can also dynamically adjust the access to the detection module based on the results of vehicle speed calculations. frequency, and transmits all information to the server 20 through the transmission module for calculation. On the other hand, the transmission module control unit 146 can also dynamically adjust the information transmission speed of the transmission module based on the vehicle speed calculation result; for example, control the transmission module to transmit when the vehicle speed is relatively fast (such as greater than 60 km/h). More information, so that the status judgment module in the backend server 20 can have enough time and information to judge abnormalities in driving behavior.

The transmission module 16 is connected to the detection module 12 and the processing module 14, and includes an information transceiver unit 162 and a communication control unit 164. Specifically, the communication control unit 164 controls the establishment and closing of the network communication connection between the electronic device 10 and the server 20; the communication method adopts but is not limited to the fourth generation mobile communication technology (4th-Generation, 4G), Fifth generation mobile communication technology (5th-Generation, 5G) or General Packet Radio Service (GPRS). The fourth and fifth generation mobile communication technologies are mobile communication technologies that support high-speed data transmission and can transmit voice and information at the same time. Their representative feature is the provision of high-speed information services; General Packet Radio Service (GPRS) has low cost and high information transmission rate. , and the packet-switched transmission mode is introduced, the user only occupies resources during the period of sending or receiving data. On the other hand, the information transceiver unit 162 includes an image transceiver unit and a text transceiver unit. The image transceiver unit automatically adjusts the transmission rate of the image information according to the speed of the vehicle so that the server 20 continues to receive images; the text transceiver unit transmits The motion status, positioning, date and time, vehicle speed and other information also enable the server 20 to continue to receive this information.

The purpose of the transmission module 16 is to transmit real-time image information to the server 20 according to the vehicle speed; in the server, the CNN (Convolutional Neural Network) model of artificial intelligence deep learning is quoted to identify the numbers in the image. Chi. In this embodiment, the current design specifications for signs prohibiting left turns and prohibiting turns on the road are circular in shape and 60-70 cm in diameter.

x可推算出可辨識號誌的最短攝像距離x為2251.67公分。 During the image recognition process of the server 20 (sign identification unit 224), illegal signs on the road should have a specific size after image detection, so that the artificial intelligence algorithm of the server 20 can identify them. For example, the image size for artificial intelligence training is 16 pixels in length and width. The viewing angle width that the camera device can capture is 120 degrees. The width and height of the captured image are 1920 and 1080 pixels; consider In the real environment, the size of illegal signs is 60-70 centimeters. When the camera device captures the image, it must be greater than or equal to 16 pixels. The schematic diagram of the camera device is as shown in the second picture. If the image is taken at 1920 * 1080, the width of 1 pixel in reality is 65/16 = 4.0625 cm. It can be estimated that the 1920 pixel is 4.0625 x 1920 = 7800 cm. According to the schematic diagram, it is assumed that the angle of the camera device is 120 degrees. , the width of the camera screen in reality is y centimeters, divide 7800 by 2

x can be deduced that the shortest camera distance x for identifiable signals is 2251.67 cm.

Furthermore, considering the demand for real-time image recognition during vehicle driving, the transmission of image information must be based on the speed of the vehicle. When the vehicle speed is high, more images are transmitted so that the signal recognition can be successfully completed. Assuming that the shortest camera distance of identifiable signs is x centimeters and the vehicle speed is s centimeters/second, the frequency of image capture of identifiable signs is 1/(x/s) images per second, and its unit is frame per second (fps), and then set the network communication speed according to the practical plan, the unit is send frame per second (sfps).

Please refer to Table 1 below. In this embodiment, the vehicle speed, camera resolution, and the size of the pre-artificial intelligence training image are used as variables to calculate the image transmission rate. Among them, the vehicle speed is set to 36, 54 and 72 kilometers per hour, the camera resolution is 1920x1080 and 720x480, and the initial artificial intelligence image training size is 64x64, 32x32, 16x16, 8x8. Under these conditions, various configurations must be transmitted every second The image planning is as follows in Table 1:

Next, the backend server 20 connected to the electronic device 10 includes: a status judgment module 22, an event processing module 24 and a database 26. In this embodiment, the server 20 performs signal recognition, image processing and area determination on the information sent by the electronic device 10. When a violation event defined in the event processing module or database occurs, the backend server uses Google's The FCM or Mail service notifies the user immediately; the server 20 may be a cloud server. Cloud server means Internet-based By using the system provided by the service provider for calculation and analysis, the shared software and hardware information can be provided to various terminal electronic devices or other devices as required. Among them, cloud computing servers can include but are not limited to SaaS (Software as a Service, Software as a Service) servers, IaaS (Infrastructure as a Service, Infrastructure as a Service) servers, PaaS (Platform as a Service, Platform as a service) servers and DaaS (Data as a service) server, and this embodiment uses a SaaS server as the cloud computing server in the system.

The characteristic of SaaS servers is that they use cloud-based centralized hosting software (or applications) and related information. The software only needs to be used through the Internet without downloading or installation. The client usually uses network transmission. The agreement is to access software and services via a browser. Thus, the software (or application) itself is not downloaded or installed on the user's electronic device, but is stored in the cloud or server provided by the service provider. Compared with traditional software that must be downloaded and installed after purchase, SaaS servers only require users to rent the software for online use, which reduces the user's purchase risk and does not require restrictions on downloading software or device requirements; for service providers, it is also Reduces the information technology cost of software and hardware architecture.

Furthermore, the state determination module 22 further includes a state determination unit 222 that determines the movement state of the vehicle; a signal recognition unit 224 that recognizes whether the image contains a traffic signal; and a signal recognition unit 224 that recognizes whether the vehicle is entering. The lane recognition unit 226 of the auxiliary lane; and the violation detection unit 228 that determines whether the vehicle is driving illegally or violates the intersection clearance rules. In the state judgment unit 222, the main states of the vehicle such as driving, idling, violation, and disconnection can be detected through the information transmitted by the electronic device 10; violations can be further subdivided into illegal turns, illegal left (right) turns, and intersections. Clearance violations, vehicle zone violations and speeding violations. In this embodiment, the detection and judgment of the status judgment unit are defined as follows:

A. Driving: When the vehicle's driving speed is greater than 0 and less than the specified speed limit, it means the vehicle is driving.

B. Idle speed: When the vehicle speed is 0 and the engine is starting, it means the vehicle is idling.

C. Violation: It is divided into the following five types of violation anomaly detection:

1. Violation reversal: Please refer to the description of the violation detection unit in the later section.

2. Illegal left (right turn): Please refer to the description of the violation detection unit in the later section.

3. Intersection clearance violation: Please refer to the description of the violation detection unit in the later section.

4. Vehicle area violation: allows users to define a circular or polygonal area on the map, and can set the attribute of the area to be prohibited from entering or prohibited from leaving. When the vehicle leaves a prohibited area or the vehicle enters a prohibited area area, it will be regarded as a vehicle area violation, and a vehicle area violation event will be triggered.

5. Speeding violation: Among the recent m pieces of data, if n items exceed the speed limit set by the system, it will be regarded as a vehicle speeding violation, and a speeding violation event will be triggered.

D. Interruption: The server has not received the information transmitted by the electronic device for more than a predetermined time. The interruption may be caused by a vehicle passing through a tunnel, a vehicle driving into a basement, or an abnormality in the electronic device.

In this embodiment, the sign identification unit 224 uses artificial intelligence deep convolutional neural network (CNN) to identify images. Image samples are first input into the neural network for training. After training, a model is generated, and then the model is used. Identify the sign. The images used for training the model must be pre-processed first. Please refer to the third figure, which is a pre-processing flow chart for processing the input images of the artificial intelligence model in this embodiment. The detailed description is as follows:

1. RGB->HSV: Convert images in RGB (red, green, blue) color space to HSV (hue, saturation, brightness), thereby allowing subsequent image processing programs to process images more sensitively.

2. Find the red area of interest (Region of Interest, ROI): After the image is converted, in the HSV color space, the red values in S and V are close to 100, and the H part is very Low. By setting a numerical filtering threshold, you can quickly find reddish areas in the image.

3. Fuzzy denoising: Fuzzy denoising is a method of blurring the image to remove noise and details in image processing. For example, the Gaussian filter is a filter that eliminates the high-frequency parts of the image.

4. Binarization: Binarization is a method of image segmentation. Binarization can convert a grayscale image into a binary image. The grayscale of pixels greater than a specific critical grayscale value is set as the grayscale maximum value; the pixel grayscale smaller than this value is set as the grayscale minimum value. Borrow This results in a binarized image.

5. Edge detection: Edge detection is a feature-searching program that can find blocks in the image for further processing.

6. Hough circle: The main function of Hough circle conversion is to find the circles in the image. The principle is to draw a circle at each edge, and then superimpose all the drawn line segments. The brightest point is the center of the circle, so you can search for circles in the shape.

7. Variable sealing gap: The purpose of the process of transforming the gap is to repair the circle detected in the previous step to produce a circular shape that is more consistent with the signal detection, so as to find the image. candidate areas in .

8. Cut the image: After all the previous processing, the circular area in the image has basically been found, so this step requires cutting the area.

9. Output signal map: After cutting the image, the signal area that needs to be identified can be output.

10. Image screening: The purpose of image screening is to screen out images that may cause misjudgment. For example, the same red circle pattern is not a traffic signal, advertising light board, installation art or various similar outdoor scenes.

11. CNN model: This embodiment uses a CNN model to train the image model. The final model will be able to detect the required left (right) turns and turns prohibited, thereby identifying traffic signs.

In the above image recognition procedure, the purpose of pre-processing is to establish a model. When the model is built, the subsequent images input by the electronic device will be processed using the trained model to determine how similar the input images are to the No Left (Right) Turn and No Turn Traffic Signs that need to be identified in the process. If the similarity is high, It means that the recognition is successful; if the level is low, it means that the image is not an illegal sign.

In this embodiment, the lane recognition unit 226 uses image processing to identify the left (right) turn auxiliary lane. Please refer to Figure 4, which is a flow chart of the lane recognition method according to the preferred embodiment of the invention. The detailed description is as follows:

1. RGB->GRAY: Convert the RGB of the original image into a grayscale image. The principle is to convert RGB into grayscale at different ratios. The grayscale conversion formula is as follows:

Gray =R*0.299+G*0.587+B*0.114

Gray =(R*299+G*587+B*114)/1000

2. Expansion type transformation (expansion): The inflation type variation considers two sets in space, the A set and the B set. When the A set is expanded by the B set, it can be expressed as A+B, A is the input image, and B is Structural element; when more than one of the input pixel and its surrounding pixels is 255 relative to the pixel value of the structural element of 1, the value of the input pixel is set to 255. The operation result will make the image look expanded. Its application is to To fill the gaps, use appropriate structural elements to fill the gaps. In this embodiment, the purpose of expansion is to connect the lane lines of the left turn auxiliary lane for further processing.

3. Contraction type change (erosion): Consider two sets in space, A set and B set. When A set is eroded by B set, it can be expressed as A-B, A is the input image, B is the structural element; when the input pixel The values of its surrounding pixels relative to the pixels whose structural element is 1 are all 255, then the value of the input pixel is set to 255, and the result of the operation will make the image appear to shrink. Erosion can remove unnecessary elements using appropriate structural elements. In this embodiment, the purpose of shrinkage is to The lane lines of the connected left-turn auxiliary lanes can be made more obvious as lines to facilitate further processing.

4. Perspective transformation (bird's eye view): Perspective transformation converts the image from forward illumination to top down illumination. The principle is to set the position coordinates of the image before and after conversion through the conversion of the coordinate system, so that the image to be detected The auxiliary lane is more visible in the image.

5. Hough line segment: Hough line segment conversion is one of the three major applications of Hough conversion. The purpose of Hough line segment conversion is to detect line segments in images. The principle is to use the voting step in the Hough line segment algorithm to achieve complex results in Find the parameters of the graph in the parameter space, and learn the shape of the edge from the parameters. In this embodiment, the application of the Hough line segment is to find the left turn auxiliary lane in the image.

6. Take a point on the left and right of the two lines, find the slope and whether the output is a curve waiting area: After the Hough line segment processing is completed, the line segments in the image have basically been filtered. In this embodiment, points are taken on two lines of the lane and their slopes are calculated. If the slope remains unchanged, it means that the vehicle is currently on the straight road; if the slope changes, it means that the vehicle is currently in the left turn auxiliary lane, and it can be output whether the current vehicle is in the left turn auxiliary lane. Judgment results in the left turn auxiliary lane.

Finally, the purpose of the violation detection unit 228 in this embodiment is to determine whether the vehicle is driving illegally or violates intersection clearance rules. First, the illegal driving behavior in this embodiment includes illegal left turns and illegal turns. Other possible implementation modes also include the detection of illegal right turns. Specifically, this embodiment uses the vehicle motion status information (such as acceleration angle) detected by the status detection unit, the positioning information (such as longitude and latitude) obtained by the positioning unit, and the signal identification results processed by the aforementioned signal identification unit. , determine whether the vehicle makes an illegal left turn or illegal turn. Please refer to Figure 5, which is a flow chart of violation detection in a preferred embodiment of the present invention. When the vehicle is turning left, the acceleration angle measured by the detection unit is less than 0; conversely, when the vehicle is turning right, The acceleration angle is greater than 0.

Continuing, the acceleration angle measured by the detection unit is represented by D, the counter with D less than 0 is represented by DC _minus , the minimum threshold value for judging the vehicle turning is represented by thr, the longitude and latitude are represented by (x, y), and the longitude and latitude at any time is represented by x(t), y(t) are represented, and the distance is represented by d. When the violation detection unit starts, DC _minus =0. When the violation detection unit detects D<0, it will add DC _minus +1 as time goes by. If the system continues to detect D<0, DC _minus will continue to increase; if DC _minus >thr is found during the increment, it will confirm whether x(tn), y(tn) has been stored. If it has not been stored yet, store it first, and then calculate the distance d ₀ between the two points based on the longitude and latitude x(t), y(t) and x(tn), y(tn) of the current time; if the next time point t+1 is also In line with D<0, DC _minus >thx, x(tn), y(tn) has been stored at this time, and x(t+1), y(t+1) and x(tn), y(tn) are calculated distance d ₁ . At the next time point t+2, if it also meets D<0, DC _minus >thr, calculate the distance d ₂ between x(t+2), y(t+2) and x(tn), y(tn); according to This program calculates all the distance information of d _t+i ; where i=0 to j, and then compares the relationship between d _t , d _t+1 , d _t+2... d _t+j . If the distance is increasing, it means that the vehicle is getting farther and farther away from the longitude and latitude x (tn), y (tn), and it is determined to be a left turn; if the distance is decreasing, it means that the vehicle is getting closer and closer to the longitude and latitude x (tn), y (tn) , judged as rotation.

When any behavior of left turn or left turn occurs, it is necessary to retrieve all the image information records of the previous n seconds to determine whether there is a sign prohibiting left turn or turn in the vehicle's driving path; when the corresponding process is detected When the signal prohibiting left turn or rotation is received, the event is triggered and the previously stored x(t-n), y(t-n) is cleared.

On the other hand, please refer to Figure 6, which is an algorithm flow chart for intersection clearance violation detection according to a preferred embodiment of the present invention. The algorithm for violating intersection clearance rules will be based on the electronic fence areas marked at important intersections. When a vehicle stays in the electronic fence area for a period of time, it will be regarded as the vehicle's staying behavior, and then it will start to detect whether the vehicle has stopped in accordance with the regulations. It is equal to the waiting area of the left-turn auxiliary lane; if the vehicle is not in the waiting area of the left-turn auxiliary lane, it indicates a violation.

The algorithm for detecting clearance violations at intersections needs to consider three variables: whether the vehicle enters the intersection area, whether the vehicle is in the left-turn auxiliary lane, and whether the vehicle is idling.

Among the three variables, whether the vehicle enters the intersection area is represented by Geo-in _int . Geo-in _int =0 indicates that the vehicle is not in the intersection area; conversely, Geo-in _int =1 indicates that the vehicle is in the intersection area. To detect whether a vehicle has entered the intersection area, this embodiment uses an electronic geographical map system to detect. In the electronic geographical map system, the actual geographical location is marked on a two-dimensional plane with longitude and latitude information. First, the relevant intersection area is pre-established in the electronic geographical map system, and then the geofence function provided by PostGIS is used to detect in real time to obtain information (Geo-in _int ) on whether the current vehicle location is in the intersection area.

To detect whether the vehicle is in the left-turn auxiliary lane, please refer to the description of the lane recognition unit mentioned above. This step refers to the calculation result of the algorithm and is represented by L-turn Lane. L-turn Lane=1 means that the vehicle is in the left-turn auxiliary lane, and L-turn Lane=0 means that the vehicle is outside the left-turn auxiliary lane. Whether the vehicle is idling can refer to the operation result of the state judgment unit. Idling is defined as the vehicle speed is 0 and the engine is starting, which is expressed as idle=1, and a moving vehicle is expressed as idle=0.

The status transfer of intersection violation detection is based on the calculation results of the above three variables: whether the vehicle enters the intersection area, whether the vehicle is in the left-turn auxiliary lane, and whether the vehicle is in an idling state. As shown in the sixth figure, when the vehicle starts driving, it is the initial state Init. Assume that the vehicle is not in the intersection detection area at this time (Geo-in _int = 0), and the vehicle moves with it; if the vehicle stops, the vehicle is At idle speed (idle=1), the vehicle status is converted from Init to Idle(1); if the vehicle is still not moving at the next time point (idle=1), the vehicle status is converted from Idle(1) to Idle(2). According to this logic, when the vehicle stays n-1 times, the state enters Idle(n-1), as shown in the figure Stay n-1 state; conversely, if the vehicle transfers from Idle(1) to Idle(n-1), If the vehicle moves (idle=0), the vehicle state will return to the initial state Init, as shown in the state transition trans 1.

When the vehicle status enters Idle(n-1) (Stay n-1 status in the figure), if the vehicle is in the intersection area (Geo-in _int =1), the vehicle is idling (idle=1) and the vehicle is on the left Turning into the auxiliary lane (L-turn Lane=1) means that the vehicle waiting to turn left in the left-turn auxiliary lane of the intersection is not an intersection clearance violation, and the status is transferred from Idle(n-1) to Idle(nL), as shown in Figure Blue StatusIdle(nL). Among them, the identification of L-turn Lane=1 is to continuously capture images to determine the status of the L-turn Lane after the vehicle enters the intersection area.

When the vehicle status enters Idle(n-1) (Stay n-1 status in the figure), if the vehicle is in the intersection area (Geo-in _int =1), the vehicle is idling (idle=1) and the vehicle is not in In the left-turn auxiliary lane (L-turn Lane=0), it means that the vehicle is parked at the intersection and is not in the area of the left-turn auxiliary lane, that is, it violates the intersection clearance rules, and the status changes from Idle(n-1) to Idle(nI) , as shown in the figure Red state Idle (nI), at this time the intersection clearance violation event is triggered. Among them, the identification of L-turn Lane=1 is to continuously capture images to determine the status of the L-turn Lane after the vehicle enters the intersection area.

On the other hand, when the vehicle is in the Blue state Idle (nL) or Red state Idle (nI), if the vehicle is no longer in the intersection area (Geo-in _int = 0), it means that the vehicle has left the intersection area, and then returns to The initial state is as shown in the figure: state transition trans 2.

Finally, the event processing module connected to the status judgment module can define at least one event based on the judgment result of the status judgment module, and send an event notification to the electronic device to notify the driver, and at the same time store these events in the data in the library. For the definition of events, please refer to Table 2 below:

Furthermore, event storage is a mechanism for storing event information in a database; the event notification mechanism is to connect externally to two services, Google Gmail or Google FCM, and notify users via email or Firebase Cloud Message.

In addition, the vehicle detection system of this embodiment can also be compatible with external systems for assistance. Google Gmail (used when events are notified by email) or Google FCM (used when events are sent to mobile phones through Google Firebase Cloud Message service) are mainly used. It is a mechanism that provides event notification, while GIS Service provides electronic geographical map services, such as GIS services such as Google Maps or OpenStreepMaps (used for event geographical location query).

In this embodiment, the electronic device further includes an application program, which includes a user interface, a vehicle real-time monitoring interface, an event query interface, a vehicle historical trajectory interface, and a system setting interface. User interface planning uses Responsive Web Design (RWD) to plan and implement the system user interface, so that a single system user interface can provide different device resolutions such as smartphones, tablets, and personal computers. user interface. After logging in to the system, a menu button is displayed. After clicking the menu button, a menu will be displayed, allowing the user to further select other functions. The menu provides functions such as map monitoring, summary monitoring, event query, historical track, system settings, and system logout.

Among them, the real-time monitoring of vehicles will use an electronic geographic map system to display the current real-time location and vehicle status of the vehicle. The electronic geographic map system (Graph Information System, GIS) will be implemented by Google Maps or Open Street Maps. The vehicle status Divided into driving, idling, violation and interruption. What's more, the violation status can be further subdivided into illegal turns, illegal left (right) turns, intersection clearance violations, vehicle area violations and speeding violations.

Event query provides users with the function of viewing and querying. "View" can provide real-time event information of the day. Click on any event to display event details; "Query" provides historical event query by date and time. The function of historical track includes historical track query and historical track playback. Historical track query will retrieve the historical track of the specified vehicle based on the start date and time, end date and time and vehicle number, and display it in the form of an electronic map; the historical track playback part, It provides users with three functions: play, stop and return to the original point. The system settings provide the company's basic information, speed limit and vehicle area settings, while the logout allows users to log out of the application.

However, the above are only preferred embodiments of the present invention, and should not be used to limit the scope of the present invention. That is, simple changes and modifications made based on the patent application scope and description content of the present invention still belong to the present invention. within the scope covered.

1:Detection system

10: Electronic devices

12:Detection module

122:Image detection unit

124: Status detection unit

126: Positioning unit

14: Processing module

142:Computing unit

144: Detection module control unit

146:Transmission module control unit

16:Transmission module

162: Information sending and receiving unit

164: Communication control unit

20:Server

22: Status judgment module

222: Status judgment unit

224: Number identification unit

226: Lane recognition unit

228: Violation detection unit

24:Event processing module

26:Database

Claims (5)

A detection system for vehicles, including: an electronic device, including, a detection module, including: an image detection unit to obtain at least one image information; a state detection unit to obtain a movement of the vehicle status information; and a positioning unit to obtain positioning information of the vehicle; a processing module connected to the detection module, the processing module including: a calculation unit to calculate the speed of the vehicle based on the positioning information; and A module control unit includes a detection module control unit and a transmission module control unit; and a transmission module is connected to the detection module and the processing module, and the transmission module includes an image transceiver unit and a text transceiver unit, the image transceiver unit transmits the at least one image information, the text transceiver unit transmits the motion status information and the positioning information; wherein, the detection module control unit regularly stores the at least one image information, the motion status information and the positioning information, the transmission module control unit dynamically controls the transmission rate of the transmission module according to the speed of the vehicle; wherein, the dynamic control means that the rate at which the transmission module transmits the at least one image information changes with the speed of the vehicle. The vehicle speed increases; and, a server is connected to the electronic device, the server includes: a state judgment module, including, a state judgment unit that judges the motion state of the vehicle based on the motion state information; A signal identification unit that identifies whether the at least one image information includes a traffic signal, wherein the signal identification unit uses an artificial intelligence deep convolutional neural network (Convolutional Neural Network, CNN) technology to identify the at least one image information; A lane recognition unit, based on the at least one image information, identifies whether the vehicle has entered the auxiliary lane; and a violation detection unit, based on the motion state of the vehicle and the at least one image information, determines whether the vehicle is driving illegally; an event A processing module is connected to the status judgment module, and the event processing module defines at least one event based on the judgment result of the status judgment module; and a database is connected to the event processing module, and the database stores The at least one event. The detection system of claim 1, wherein the transmission module further includes a communication control unit to control a connection switch between the electronic device and the server. The detection system as described in claim 1, wherein the electronic device further includes an application program, the application program includes a user interface, a vehicle real-time monitoring interface, an event query interface, a vehicle historical trajectory interface and a system setting interface. The detection system as described in claim 1, wherein the artificial intelligence deep convolutional neural network technology uses the at least one image information as a sample to input into the neural network for training, generates a model after training, and then uses the model to identify the at least one Whether the image information contains traffic signals. A violation detection method for vehicles, including: a. Determining whether the vehicle is located in an intersection area based on a geographical map system; b. Determining whether the vehicle is in an idling state based on the speed of the vehicle; c. Real-timely detect at least one image information of the vehicle's driving path, and determine whether the vehicle is driving in the general lane or the auxiliary lane based on the at least one image information. The determination method includes: c1. Convert the RGB image of the at least one image information It is a grayscale image; c2. Through the application of expansion transformation, the lane lines of one lane in the at least one image information can be connected; c3. Through the application of contraction transformation, the connected lane lines can clearly form a line; c4. Through Apply perspective transformation to make the detected lane appear clearly in the at least one image; c5. Identify the lane in the at least one image through the application of Hough line segments; c6. Take at least one point on two lines of the lane and find the A slope of at least one point. If the slope remains unchanged, it means that the vehicle is driving in the general lane; if the slope changes, it means that the vehicle is driving in the auxiliary lane; and, d. Based on whether the vehicle is located in the intersection area and whether the vehicle is idling. status and whether the vehicle enters the auxiliary lane to determine whether the vehicle violates the regulations.