Six brand new applications of the city license plate recognition system

License plate recognition technology is an application of computer video image recognition technology in vehicle license plate recognition. It requires that the license plate can be extracted and identified from complex backgrounds through license plate extraction, image preprocessing, feature extraction and license plate character recognition. Other technologies, such as identification of vehicle brands, currently have a recognition rate of 96% for letters and numbers, and a recognition rate of 95% for Chinese characters. With the development, license plate recognition technology has strengthened the use of six new technologies.

1, high-performance embedded processing system

In the traditional high-definition system, the system sends the pictures taken by the camera to the processing systems such as IPCs for identification processing and storage operations. The requirements for transmission and processing equipment are very high, and it is difficult to meet the design processing capability requirements of the system.

The new intelligent HD camera embedded processing solution can effectively solve the above problems. Each lane is equipped with a 2 million-pixel smart HD camera with a resolution of up to 1600 x 1200 pixels. It is responsible for the capture of facial features, automatic license plate recognition, and auto color recognition of the front passengers of the vehicle. The processing result is transmitted through the ether. The network is sent to the smart HD camera terminal server.

The terminal server adopts an embedded linux dedicated system to receive, store, and upload vehicle traffic information. At the same time, the user can view and manage all front-end devices through the terminal server. Each intelligent high-definition camera terminal server can simultaneously connect up to 20 smart HD cameras. Greatly improve system processing performance.

2. Integrated design of image acquisition and processing

The intelligent HD camcorder integrates an industrial-grade progressive scan CCD and a high-speed DSP chip to independently complete functions such as image capture, license plate recognition, car body color recognition, and vehicle speed acquisition. After the vehicle information is acquired, the result assembly and communication module is packaged and uploaded via Ethernet. The terminal server enhances system processing capabilities, making the system structure clear and concise, and easy to expand.

3, feedback control of all-weather high-definition imaging

The entire imaging system is a sophisticated system consisting of smart HD cameras, smart fill lamps and imaging control software. The orderly cooperation and feedback control between them makes the images of vehicles captured during the day and night high in resolution, ensuring the body, license plates and The facial features of the front passengers in the vehicle are all clearly visible.

The system integrates the reflection characteristics of the front windshield of the vehicle on the light, the condition of the film, and the ambient light exposure. It adopts a special lens, special imaging control strategy and fill light, and arranges a reasonable device layout method, making the system all-weather. All types of models can effectively solve problems such as front windshield reflection and direct light, ensuring that the facial features of the body, license plates, and front passengers of the vehicle are clearly identifiable.

4, car color recognition

It can automatically identify the depth of the car's colors and the colors of 10 common car bodies, and provide new scientific and technological means for public security inspections and detection of criminal cases.

The system can automatically distinguish whether the vehicle is a dark or light-colored vehicle; and identify 10 common vehicle colors, including white, silver, gray (with gray and silver gray), black, red (red and dark red), dark blue , blue (blue and blue), yellow (golden and yellow), green (green and dark green), brown (light brown and brown), colors other than 10 colors (including colored vehicles) belong to other colors .

The classification accuracy of depth and depth is not less than 80%; the recognition accuracy of 10 kinds of common car body colors is not less than 60%.

5, video-assisted triggering, auxiliary speed

Video triggering is used as an auxiliary trigger. In the event of a failure of the coil trigger, it can alarm in a short time and replace the coil to trigger. In this way, the user can obtain the fault information in the first time, and can still effectively monitor the road operation status before the problem is solved, and can achieve continuous monitoring and timely response.

6. Front-end and back-end independent proprietary network structures

Two independent network structures are adopted, namely, the network between the terminal server and the intelligent HD camera and the network between the terminal server and the center are independent. In this way, the occupation of the IP address of the public security network can be reduced, the increase of equipment or IP change within the monitoring point is more convenient, and the network security is also greatly improved.