introduction
Intelligent Transportation System (ITS) is an important component of safe cities and smart cities. It is rising and accelerating in major cities. The main target of ITS management is vehicles. Under the current vehicle control system, the license plate number is the logo. The unique identification of the vehicle (Note: Although RFID and other technologies can also identify the vehicle, but it has not been promoted to the entire vehicle management system, and the vehicle is required to be modified, currently not universally applicable), the license plate automatic identification technology can be used without any changes in the vehicle. Under the circumstances, automatic registration and verification of vehicle identities is realized, so the license plate recognition system is one of the important components of modern intelligent transportation systems.
The main application of the license plate recognition system in intelligent transportation is embodied in the intelligent monitoring records and illegal automatic recording of road vehicles (including red light, ban left, pressure line, retrograde, overspeed, etc.), which are used to record the number plate of the vehicle and index it. Vehicle records and owner files, traffic management enforcement, and can be derived from false deck applications. In addition, license plate recognition technology can also be applied to various occasions such as highway toll collection, parking management, weighing system, traffic guidance, highway inspection, vehicle scheduling, and vehicle detection.
The license plate recognition system is widely used and will face various scenarios and special situations. How license plate recognition technology can better serve these applications and whether it faces new problems, we start with how the license plate recognition system works.
License plate recognition system works
The license plate recognition is a pattern recognition technology that uses the dynamic video or static image of the vehicle to automatically identify the license plate number and license plate color. The core technologies include license plate location algorithm, license plate character segmentation algorithm and optical character recognition algorithm. A complete license plate recognition system should include vehicle detection, image acquisition, license plate recognition and other parts. When the vehicle detection section detects the arrival of the vehicle, the image acquisition unit is triggered to collect the current video image. The license plate recognition unit processes the image, locates the position of the license plate, and then separates the characters in the license plate to recognize it and then composes the license plate number.
The vehicle detection part usually adopts the sense coil or radar, and some license plate recognition systems also have the function of judging whether there is a car through the video image, which is called video vehicle detection.
Since the roads are open 24 hours a day, the license plate recognition system needs to work all day and all-weather. In order to ensure the accuracy of identification at night, an LED strobe light or flash can be used to fill the light.
The structure of a typical license plate recognition system is shown in the figure below. The front-end equipment is connected to the back-end platform through the transmission network.
The license plate recognition system usually completes the identification output work through the following steps:
Vehicle detection: It can detect the passing of the vehicle by means of buried coil detection, infrared detection, radar detection technology, video detection and other methods, and trigger image capture and capture.
Image acquisition: Real-time, continuous recording and acquisition of passing vehicles through high-definition video capture.
Preprocessing: noise filtering, auto white balance, auto exposure and gamma correction, edge enhancement, contrast adjustment, etc.
License Plate Positioning: Rows are scanned on the grayscale image after image preprocessing to determine the license plate area.
Character segmentation: After the license plate area is located in the image, the character area is accurately positioned through processing such as graying and binarization, and then character segmentation is performed according to the character size feature.
Character recognition: The segmented characters are scaled, extracted, and matched with the standard character representation in the character database template.
Result output: The license plate recognition result is output in text format.
License plate recognition system function and application
The main functions of the license plate recognition system in the current market include:
1) The vehicle license plate is automatically identified. The information includes complete license plate information, colors, characters, Chinese characters, and numbers;
2) Automatic detection of speed;
3) The identification and warning of illegal black cards;
4) Linkage control of vehicle identification information and vehicle information in the vehicle management office;
5) Vehicle direction judgment monitoring.
The main applications of the license plate recognition system in the current market include:
1) Intelligent traffic management at traffic intersections;
2) Automatic collection of traffic information;
3) The police and other law enforcement agencies set up temporary inspection stations to conduct inspections of passing vehicles and prioritize the identification of vehicles to be investigated;
4) Automatic charging system for bayonet, tunnel and other bayonet;
5) Car entrance and exit management in modern residential quarters, parking lots, and important authorities;
6) Road security bayonet snapping recognition, traffic monitoring.
License Plate Recognition System Solution
There are usually three types of license plate recognition system implementations. One is the capture camera + IPC, which is the earliest application of a program. The camera image (coil, radar trigger or video detection) captures the vehicle image, and the IPC software identifies the license plate. An industrial computer can manage multiple cameras (multi-lane) at the same time. Second, it captures the camera and the embedded analysis host. The industrial computer with lower reliability is changed to the embedded host, and the capture camera does not change, so this method Has gradually replaced the first scheme; third is embedded integrated capture camera, set capture, control, identification, video, compression, transmission in one, greatly simplifies the management of the terminal equipment and the back-end platform, at the same time in the reliability and security Sex, installation and maintenance convenience, and environmental adaptability have all been improved and are becoming the most promising implementation solutions.
Here we talk about the current high-definition capture camera that integrates vehicle video detection and license plate recognition on the market. Compared to traditional coil triggering methods, video detection can avoid damage to the road surface, no external detection equipment, no need to correct the trigger position, Savings can be more flexible in detecting vehicle behavior (such as illegal U-turn, pressure line, left turn, etc.), and more suitable for mobile, portable applications. However, the vehicle license plate recognition system needs to perform video vehicle detection, requires high processing speed, and adopts an excellent algorithm to achieve image acquisition and processing without any frame loss. If the processing speed is slow, it will result in dropped frames, making it impossible for the system to detect a vehicle with a fast traveling speed. It is also difficult to ensure that recognition processing is started at a position that is conducive to recognition and the system recognition rate is affected. Therefore, the combination of video vehicle detection and automatic license plate recognition has a certain degree of technical difficulty, which is why there are not many products currently appearing.
In addition to the front-end camera and analysis host, a license plate recognition system should also have a corresponding back-end management system, and it affects whether this license plate recognition system is easy to use. The functions of the back-end management system usually include:
1.Reliable storage of recognition results and vehicle image data, when the multi-functional system operation makes the network go wrong, it can protect the image data from being lost, at the same time, it is convenient for manual inspection afterwards;
2. Effective automatic comparison and query techniques, the identified license plate number shall be automatically compared with the thousands of license plate numbers in the database and the alarm shall be prompted. If the license plate number is not correctly read, fuzzy query shall be used. Technology can produce relatively "best" comparisons;
3. For networking operation, it provides functions such as real-time communication, network security, remote maintenance, dynamic data exchange, automatic database update, hardware parameter setting, and system fault diagnosis.
Evaluation of License Plate Recognition System
The Ministry of Supervision of the People's Republic of China has clear technical specifications and management regulations for the monitoring of road traffic and vehicle movements, mainly including the "General Technical Specifications for Automatic Red-light Recording Systems" (GA/T496-2009) and "General Technical Conditions for Intelligent Monitoring and Recording Systems for Road Vehicles" ( GA/T497-2009), "Technical Specifications for Imagery Forensics of Road Traffic Safety Illegal Behavior" (GA/T832-2009), and "Automatic Number Identification Technical Specifications for Motor Vehicles" (GA/T833-2009).
From the technical evaluation of a license plate recognition system, there are two main indicators, namely the recognition accuracy rate and the recognition speed.
· Recognition accuracy
A license plate recognition system is practical, the most important indicator is the recognition rate and recognition accuracy. According to the definition in GA/T497-2004 and GA/T497-2009:
Recognition rate = The total number of vehicles whose number plate/number plate information is automatically recognized.
Recognition Accuracy = Number Plate Information Identify the correct number of vehicles/number plate number of valid vehicles.
(Note: The effective number plate information means that the vehicle number plate is complete, clear, and the installation specifications, and there is no shelter or contamination.)
In the newly issued GA/T497-2009, the number plate recognition rate index evaluation was canceled, only the plate number recognition accuracy rate was retained, and the accuracy rate of day plate number plate recognition should not be less than 90%; the night plate number plate recognition accuracy rate Should not be less than 80%.
Since the conclusion that the number plate information is valid requires manual determination, it is also necessary to store both the vehicle number plate image and the recognition result in order to retrieve the view. Then, quantitative statistics are made on the actually passed vehicle images and correct human recognition results to obtain intermediate results of the recognition rate, the recognition accuracy rate, and the credibility and the misrecognition rate.
In order to test the recognition rate of a license plate recognition system, it is necessary to install the system in a practical application environment, run it for more than 24 hours in all weather conditions, and collect vehicle license plates for identification of at least 1000 natural vehicles.
· Recognition speed
The recognition speed determines whether a license plate recognition system can meet real-time application requirements. A system with a high recognition rate, if it takes several seconds or even several minutes to recognize the result, then this system will have no practical significance because it can not meet the real-time requirements in practical applications. For example, one of the functions of license plate recognition in highway toll collection is to reduce travel time, which is a powerful guarantee for reducing transit time and avoiding traffic jams in this type of application.
According to the requirements in GA/T833-2009, the identification time is ≤(A/B)×(K×100)(ms).
A in the above formula represents the resolution of the image for recognition; B is a fixed constant, and its value is 768*576=442368; K is the number of license plates present in the image.
That is, when the license plate image is 768×576 pixels, when there is a number plate in the image, the recognition time is ≤100ms; when there are two number plates in the image, the recognition time is ≤200ms; when there are three number plates in the image The recognition time is ≤300ms; when there are four number plates in the image, the recognition time is ≤400ms.
License plate recognition system problems and new technology applications
From the above we can see that the higher the recognition rate and recognition accuracy of the license plate recognition system, the better, but we must clearly recognize that the recognition rate of 100% is impossible, on the one hand because the license plate is defaced, blurred, blocked, or bad The weather (snow hail and haze, etc.) will seriously affect the recognition effect. On the other hand, the segmentation and recognition of Chinese and English characters is more difficult, such as the "Ci" word error segmentation, and "0-Q", "2-Z", "4-A", "5-S", "7-T", "8-B", "OD" and other confusing characters. Because the statistics of recognition rate is based on the total number of vehicles with valid license plate information, if the vehicle license plates under various circumstances and situations are considered, the recognition rate of the license plate recognition system in practical applications will be greatly reduced. Still rely on artificial judgment, identification.
Aiming at the disadvantages of the low recognition rate of the traditional license plate recognition algorithm, a recognition method based on convolutional neural network has emerged. Through the sample learning of the license plate character images, the weight parameters of each layer of the neural network are optimized, and thus it is very large. To improve the character recognition rate of the license plate. The simulation results show that using the recognition method of convolution neural network to identify the characters in the license plate, the correct recognition rate can reach 99%, and the recognition rate and anti-interference performance are obviously better than the traditional identification methods such as structural feature method and template matching method. The latter two were only 94% and 95% respectively.
Using the advantages of neural networks, an improved recognition mechanism based on convolutional neural network is used to identify characters in license plates. The recognition method optimizes the weight parameters of each layer in the network system by learning the license plate characters under ideal pre-processing conditions, which greatly improves the character recognition rate in the license plate. In practical applications, for the disadvantages such as unclear license plate positioning and incorrect character segmentation during pre-processing, if the network structure is further optimized, license plate characters under poor pre-conditioning conditions can be identified.
Intelligent Transportation System (ITS) is an important component of safe cities and smart cities. It is rising and accelerating in major cities. The main target of ITS management is vehicles. Under the current vehicle control system, the license plate number is the logo. The unique identification of the vehicle (Note: Although RFID and other technologies can also identify the vehicle, but it has not been promoted to the entire vehicle management system, and the vehicle is required to be modified, currently not universally applicable), the license plate automatic identification technology can be used without any changes in the vehicle. Under the circumstances, automatic registration and verification of vehicle identities is realized, so the license plate recognition system is one of the important components of modern intelligent transportation systems.
The main application of the license plate recognition system in intelligent transportation is embodied in the intelligent monitoring records and illegal automatic recording of road vehicles (including red light, ban left, pressure line, retrograde, overspeed, etc.), which are used to record the number plate of the vehicle and index it. Vehicle records and owner files, traffic management enforcement, and can be derived from false deck applications. In addition, license plate recognition technology can also be applied to various occasions such as highway toll collection, parking management, weighing system, traffic guidance, highway inspection, vehicle scheduling, and vehicle detection.
The license plate recognition system is widely used and will face various scenarios and special situations. How license plate recognition technology can better serve these applications and whether it faces new problems, we start with how the license plate recognition system works.
License plate recognition system works
The license plate recognition is a pattern recognition technology that uses the dynamic video or static image of the vehicle to automatically identify the license plate number and license plate color. The core technologies include license plate location algorithm, license plate character segmentation algorithm and optical character recognition algorithm. A complete license plate recognition system should include vehicle detection, image acquisition, license plate recognition and other parts. When the vehicle detection section detects the arrival of the vehicle, the image acquisition unit is triggered to collect the current video image. The license plate recognition unit processes the image, locates the position of the license plate, and then separates the characters in the license plate to recognize it and then composes the license plate number.
The vehicle detection part usually adopts the sense coil or radar, and some license plate recognition systems also have the function of judging whether there is a car through the video image, which is called video vehicle detection.
Since the roads are open 24 hours a day, the license plate recognition system needs to work all day and all-weather. In order to ensure the accuracy of identification at night, an LED strobe light or flash can be used to fill the light.
The structure of a typical license plate recognition system is shown in the figure below. The front-end equipment is connected to the back-end platform through the transmission network.
The license plate recognition system usually completes the identification output work through the following steps:
Vehicle detection: It can detect the passing of the vehicle by means of buried coil detection, infrared detection, radar detection technology, video detection and other methods, and trigger image capture and capture.
Image acquisition: Real-time, continuous recording and acquisition of passing vehicles through high-definition video capture.
Preprocessing: noise filtering, auto white balance, auto exposure and gamma correction, edge enhancement, contrast adjustment, etc.
License Plate Positioning: Rows are scanned on the grayscale image after image preprocessing to determine the license plate area.
Character segmentation: After the license plate area is located in the image, the character area is accurately positioned through processing such as graying and binarization, and then character segmentation is performed according to the character size feature.
Character recognition: The segmented characters are scaled, extracted, and matched with the standard character representation in the character database template.
Result output: The license plate recognition result is output in text format.
License plate recognition system function and application
The main functions of the license plate recognition system in the current market include:
1) The vehicle license plate is automatically identified. The information includes complete license plate information, colors, characters, Chinese characters, and numbers;
2) Automatic detection of speed;
3) The identification and warning of illegal black cards;
4) Linkage control of vehicle identification information and vehicle information in the vehicle management office;
5) Vehicle direction judgment monitoring.
The main applications of the license plate recognition system in the current market include:
1) Intelligent traffic management at traffic intersections;
2) Automatic collection of traffic information;
3) The police and other law enforcement agencies set up temporary inspection stations to conduct inspections of passing vehicles and prioritize the identification of vehicles to be investigated;
4) Automatic charging system for bayonet, tunnel and other bayonet;
5) Car entrance and exit management in modern residential quarters, parking lots, and important authorities;
6) Road security bayonet snapping recognition, traffic monitoring.
License Plate Recognition System Solution
There are usually three types of license plate recognition system implementations. One is the capture camera + IPC, which is the earliest application of a program. The camera image (coil, radar trigger or video detection) captures the vehicle image, and the IPC software identifies the license plate. An industrial computer can manage multiple cameras (multi-lane) at the same time. Second, it captures the camera and the embedded analysis host. The industrial computer with lower reliability is changed to the embedded host, and the capture camera does not change, so this method Has gradually replaced the first scheme; third is embedded integrated capture camera, set capture, control, identification, video, compression, transmission in one, greatly simplifies the management of the terminal equipment and the back-end platform, at the same time in the reliability and security Sex, installation and maintenance convenience, and environmental adaptability have all been improved and are becoming the most promising implementation solutions.
Here we talk about the current high-definition capture camera that integrates vehicle video detection and license plate recognition on the market. Compared to traditional coil triggering methods, video detection can avoid damage to the road surface, no external detection equipment, no need to correct the trigger position, Savings can be more flexible in detecting vehicle behavior (such as illegal U-turn, pressure line, left turn, etc.), and more suitable for mobile, portable applications. However, the vehicle license plate recognition system needs to perform video vehicle detection, requires high processing speed, and adopts an excellent algorithm to achieve image acquisition and processing without any frame loss. If the processing speed is slow, it will result in dropped frames, making it impossible for the system to detect a vehicle with a fast traveling speed. It is also difficult to ensure that recognition processing is started at a position that is conducive to recognition and the system recognition rate is affected. Therefore, the combination of video vehicle detection and automatic license plate recognition has a certain degree of technical difficulty, which is why there are not many products currently appearing.
In addition to the front-end camera and analysis host, a license plate recognition system should also have a corresponding back-end management system, and it affects whether this license plate recognition system is easy to use. The functions of the back-end management system usually include:
1.Reliable storage of recognition results and vehicle image data, when the multi-functional system operation makes the network go wrong, it can protect the image data from being lost, at the same time, it is convenient for manual inspection afterwards;
2. Effective automatic comparison and query techniques, the identified license plate number shall be automatically compared with the thousands of license plate numbers in the database and the alarm shall be prompted. If the license plate number is not correctly read, fuzzy query shall be used. Technology can produce relatively "best" comparisons;
3. For networking operation, it provides functions such as real-time communication, network security, remote maintenance, dynamic data exchange, automatic database update, hardware parameter setting, and system fault diagnosis.
Evaluation of License Plate Recognition System
The Ministry of Supervision of the People's Republic of China has clear technical specifications and management regulations for the monitoring of road traffic and vehicle movements, mainly including the "General Technical Specifications for Automatic Red-light Recording Systems" (GA/T496-2009) and "General Technical Conditions for Intelligent Monitoring and Recording Systems for Road Vehicles" ( GA/T497-2009), "Technical Specifications for Imagery Forensics of Road Traffic Safety Illegal Behavior" (GA/T832-2009), and "Automatic Number Identification Technical Specifications for Motor Vehicles" (GA/T833-2009).
From the technical evaluation of a license plate recognition system, there are two main indicators, namely the recognition accuracy rate and the recognition speed.
· Recognition accuracy
A license plate recognition system is practical, the most important indicator is the recognition rate and recognition accuracy. According to the definition in GA/T497-2004 and GA/T497-2009:
Recognition rate = The total number of vehicles whose number plate/number plate information is automatically recognized.
Recognition Accuracy = Number Plate Information Identify the correct number of vehicles/number plate number of valid vehicles.
(Note: The effective number plate information means that the vehicle number plate is complete, clear, and the installation specifications, and there is no shelter or contamination.)
In the newly issued GA/T497-2009, the number plate recognition rate index evaluation was canceled, only the plate number recognition accuracy rate was retained, and the accuracy rate of day plate number plate recognition should not be less than 90%; the night plate number plate recognition accuracy rate Should not be less than 80%.
Since the conclusion that the number plate information is valid requires manual determination, it is also necessary to store both the vehicle number plate image and the recognition result in order to retrieve the view. Then, quantitative statistics are made on the actually passed vehicle images and correct human recognition results to obtain intermediate results of the recognition rate, the recognition accuracy rate, and the credibility and the misrecognition rate.
In order to test the recognition rate of a license plate recognition system, it is necessary to install the system in a practical application environment, run it for more than 24 hours in all weather conditions, and collect vehicle license plates for identification of at least 1000 natural vehicles.
· Recognition speed
The recognition speed determines whether a license plate recognition system can meet real-time application requirements. A system with a high recognition rate, if it takes several seconds or even several minutes to recognize the result, then this system will have no practical significance because it can not meet the real-time requirements in practical applications. For example, one of the functions of license plate recognition in highway toll collection is to reduce travel time, which is a powerful guarantee for reducing transit time and avoiding traffic jams in this type of application.
According to the requirements in GA/T833-2009, the identification time is ≤(A/B)×(K×100)(ms).
A in the above formula represents the resolution of the image for recognition; B is a fixed constant, and its value is 768*576=442368; K is the number of license plates present in the image.
That is, when the license plate image is 768×576 pixels, when there is a number plate in the image, the recognition time is ≤100ms; when there are two number plates in the image, the recognition time is ≤200ms; when there are three number plates in the image The recognition time is ≤300ms; when there are four number plates in the image, the recognition time is ≤400ms.
License plate recognition system problems and new technology applications
From the above we can see that the higher the recognition rate and recognition accuracy of the license plate recognition system, the better, but we must clearly recognize that the recognition rate of 100% is impossible, on the one hand because the license plate is defaced, blurred, blocked, or bad The weather (snow hail and haze, etc.) will seriously affect the recognition effect. On the other hand, the segmentation and recognition of Chinese and English characters is more difficult, such as the "Ci" word error segmentation, and "0-Q", "2-Z", "4-A", "5-S", "7-T", "8-B", "OD" and other confusing characters. Because the statistics of recognition rate is based on the total number of vehicles with valid license plate information, if the vehicle license plates under various circumstances and situations are considered, the recognition rate of the license plate recognition system in practical applications will be greatly reduced. Still rely on artificial judgment, identification.
Aiming at the disadvantages of the low recognition rate of the traditional license plate recognition algorithm, a recognition method based on convolutional neural network has emerged. Through the sample learning of the license plate character images, the weight parameters of each layer of the neural network are optimized, and thus it is very large. To improve the character recognition rate of the license plate. The simulation results show that using the recognition method of convolution neural network to identify the characters in the license plate, the correct recognition rate can reach 99%, and the recognition rate and anti-interference performance are obviously better than the traditional identification methods such as structural feature method and template matching method. The latter two were only 94% and 95% respectively.
Using the advantages of neural networks, an improved recognition mechanism based on convolutional neural network is used to identify characters in license plates. The recognition method optimizes the weight parameters of each layer in the network system by learning the license plate characters under ideal pre-processing conditions, which greatly improves the character recognition rate in the license plate. In practical applications, for the disadvantages such as unclear license plate positioning and incorrect character segmentation during pre-processing, if the network structure is further optimized, license plate characters under poor pre-conditioning conditions can be identified.
Do you like the Car star lights?Have you ever known a supercar with a starry sky on the ceiling using fiber optic decor technology? We have successfully researched and applied this technology in China. Beautiful starry sky on the car ceiling with hundreds, thousands of stars and color changing, sparkling effects will make your car unique and class that only have in supercar.
The product uses Ceiling Star of DSPOF, suitable for the 12V available power of the car.If you wanna buy this kind of car roof star light,please contact with us.
Car Roof Light,Car Roof Star Light,Car Star Lights,Car Roof Top Light
Jiangxi Daishing POF Co.,Ltd , https://www.opticfibrelights.com