The triggering method of license plate recognition

2026-03-23 09:48:44

The license plate recognition system has two triggering methods: one is external triggering, and the other is video triggering.

The external triggering working mode refers to using coils, infrared or other detectors to detect the vehicle passing signal. After the license plate recognition system receives the vehicle triggering signal, it collects the vehicle image, automatically identifies the license plate, and performs subsequent processing. The advantage of this method is high triggering rate and stable performance; the disadvantage is that it requires cutting the ground to install coils, which results in a large construction volume.

The video triggering method refers to the license plate recognition system using dynamic motion target sequence image analysis and processing technology to detect the vehicle movement status on the lane in real time. When a vehicle passes, it captures the vehicle image, identifies the license plate, and performs subsequent processing. The video triggering method does not require any hardware vehicle detectors such as coils, infrared or others. The advantage of this method is that it is convenient for construction and does not require cutting the ground to install coils or installing vehicle detectors and other components, but its disadvantages are also very significant. Due to the limitations of the algorithm, the triggering rate and recognition rate of this scheme are much lower than those of the external triggering method.

1) Indirect method: It refers to identifying the license plate and related information by recognizing the information stored in the IC card or barcode installed on the car. The IC card technology has high recognition accuracy, reliable operation, and can work 24/7, but its entire device is expensive, the hardware equipment is very complex, and it is not suitable for off-site operations; the barcode technology has the advantages of fast recognition speed, high accuracy, strong reliability, and low cost. However, the scanner requirements are very high. In addition, both require the formulation of a national unified standard, and it is impossible to verify whether the vehicle and the barcode match, which is also a technical drawback, making it difficult to promote in a short period of time.

2) Direct method: Based on image-based license plate recognition technology belongs to the direct method and is an inductive-type intelligent license plate recognition method. It can collect and conduct real-time intelligent recognition of the license plate number of moving or stationary vehicles without any dedicated vehicle license plate signal transmission equipment. Compared with the indirect method recognition system, first, this system saves equipment installation and a large amount of funds, thereby improving economic benefits; second, due to the adoption of advanced computer application technology, it can improve the recognition speed and better solve the real-time problem; third, it is based on image recognition, so human participation can solve the recognition errors in the system, while other methods are difficult to interact with humans.

The direct method generally includes image processing technology, traditional pattern recognition technology, and artificial neural network technology.

1) Image processing technology: The research on solving license plate recognition using image processing technology began in the 1980s, but both at home and abroad only discussed a specific problem in license plate recognition and usually only used simple image processing technology to solve it, and did not form a complete system framework. The recognition process was to use industrial television cameras to take the front image of the car and then give it to the computer for simple processing, and ultimately still required human intervention. For example, the recognition of the provincial characters in the vehicle license plate, in 1985, someone using common image processing techniques proposed that the classification of Chinese character recognition was based on extracting the features of the Chinese character, selecting the floating closed value according to the projection histogram of the Chinese character, extracting the peak value in the vertical direction of the Chinese character, using the tree-shaped lookup table method for rough classification; then, according to the projection histogram in the horizontal direction of the Chinese character, selecting an appropriate closed value, performing quantization processing, forming a variable-length chain code, and using the dynamic programming method to find the minimum distance from the standard mode chain code to complete the automatic recognition of the provincial name of the Chinese character.

2) Traditional pattern recognition technology. Traditional pattern recognition technology refers to the structure feature method, statistical feature method, etc. In the 1990s, due to the development of computer vision technology, systematic research on license plate recognition began. In 1990, AS. Johnson et al. utilized computer vision technology and image processing techniques to develop an automatic vehicle license plate recognition system. This system was divided into three parts: image segmentation, feature extraction and template construction, and character recognition. By using the different histograms corresponding to different thresholds, a large number of statistical experiments were conducted to determine the threshold range of the image histogram for the license plate position, thereby segmenting the license plate based on the specific threshold corresponding to the histogram, and then using the preset standard character template for pattern matching to recognize the characters.

3) Artificial neural network technology. In recent years, some countries with developed computer and related technologies have begun to explore the use of artificial neural network technology to solve the problem of automatic license plate recognition. For example, in 1994, M. M. M. FANHY successfully applied the BAM neural network method to automatically recognize the characters on the license plate. The BAM neural network is a bidirectional associative single-layer network composed of identical neurons. Each character template corresponds to a unique BAM matrix. By comparing with the characters on the license plate, the correct license plate number is identified. The drawback of this approach using the BAM neural network method is that it fails to solve the contradiction between the storage capacity and processing speed of the recognition system.