The license plate recognition technology refers to a technology that can detect vehicles on the monitored road surface and automatically extract vehicle license plate information (including Chinese characters, English letters, Arabic numerals, and number plate colors) for processing. The license plate recognition is one of the important components in the modern intelligent transportation system and it is widely used. Based on digital image processing, pattern recognition, computer vision and other technologies, it analyzes the images or video sequences of the vehicles captured by the camera to obtain the unique license plate number of each vehicle, thereby completing the recognition process. However, for the identification of detailed road license plates, there are four major issues that need urgent solution.

License plate positioning and character segmentation

Refers to the position of the license plate in the captured image, extracts the image of the license plate, and then divides the characters in the license plate. The difficulty in localization of the license plate area mainly comes from the collected images. Due to the diversity of the license plate images collected, and when the images are collected, it is affected by many factors, such as rain, fog, and light, which makes some license plate image quality appear different degrees. The difference, under normal circumstances, the background of the collected images is very complex, the captured image is a picture collected in high-speed motion, so the location of the license plate in the picture is not fixed, the size of the license plate is not the same, all of the above interference factors are The license plate area positioning and character segmentation brought difficulties, thus affecting the license plate recognition rate.

The conflict between high resolution and recognition speed

From analog cameras to high-definition cameras, the problem of high resolution and speed of recognition of images can also be caused. The advantages of HD are self-evident, but everything is two-sided. When the license plate recognition is mainly reflected: HD pictures due to the wide coverage of the picture, multiple license plate recognition may appear in the picture at the same time. This requires a high speed for license plate recognition. The license plate recognition system may be too slow to handle high-definition video streaming because of the excessive resource consumption of the identification system. This may cause problems. Leakage phenomenon, and it is difficult to achieve the accuracy of the vehicle capture rate and license plate recognition.

The adaptability of license plate recognition system is in urgent need of strengthening

At present, our country's license plate recognition products require that the size of the identified license plate is fixed, while license plates that are too large and too small can not be accurately identified. This poses the problem that some license plates cannot be identified in the event of video triggering. In addition, in some field environments, due to the influence of external conditions, the camera can not be set in the most reasonable position, which will cause the license plate in the picture to shift to different degrees.

License plate recognition system has poor recognition of dirty license plates

In the actual application process in highways and cities, it is difficult to ensure that the involved license plates are not defaced. After a license plate is used for several years, pollution and wear will inevitably occur, and the vehicles that drive on the road are also It is difficult to guarantee that all the standard license plates are clean. Therefore, in the actual environment, in the face of damaged old license plates, how to improve the recognition capability of the license plate recognition system is also a problem that needs to be solved.