License plate recognition error _ license plate recognition can not be solved

In order to effectively curb the red light violations of vehicles in the city, the detection system of bayonet-type red light violations is now vigorously installed at various intersections in the city. The most important component of this system is the license plate recognition module. The following small series introduces you to the common problems of the license plate recognition system and its solutions.

License plate location and character segmentation

This means determining the position of the license plate in the captured image, extracting the image of the license plate, and then separating the characters in the license plate. The difficulty of positioning the license plate area is mainly from the collected images. Due to the diversity of the collected license plate images, and the image is affected by many factors, such as rain, fog, light, etc., some license plate image quality appears different degrees. The difference, in general, the background of the captured image is very complicated, the captured image is a picture captured in high-speed motion, so the position of the license plate in the picture is not fixed, the size of the license plate is not the same, the above various interference factors, It brings difficulties to the license plate area positioning and character segmentation, thus affecting the recognition rate of the license plate.

High resolution and recognition speed

From analog cameras to high-definition cameras, it also raises the problem of high resolution and recognition speed. The advantages of HD are self-evident, but everything is two-sided. When the license plate is recognized, it is mainly reflected in the fact that high-definition pictures may have multiple license plate recognitions in the picture due to the wide coverage of the pictures. This requires a high speed for the recognition of the license plate. The license plate recognition system is too large for the high-definition video stream, and the processing speed is too slow due to the excessive demand for the identification system resources, which may lead to the emergence of Leakage phenomenon, it is difficult to achieve an increase in vehicle capture rate and license plate recognition accuracy.

The adaptability of the license plate recognition system needs to be strengthened

At present, China's license plate recognition products require that the identified license plate size is fixed, and the license plates that are too large or too small are generally not accurately identified. This causes a problem that part of the license plate cannot be recognized in the case of video triggering. In addition, in some on-site environments, due to external conditions, it is impossible to mount the camera in the most reasonable position, which will cause different degrees of deviation of the license plate in the picture.

The license plate recognition system is not good for the identification of dirty license plates.

In the actual application process on the road and in the city, it is difficult to ensure that the license plates involved are not defaced. After several years of use, the license plate will inevitably become polluted and worn, and the vehicles driving on the road are also It is difficult to ensure that the standard is a clean license plate. Therefore, in the actual environment, in the face of damaged and worn license plates, how to improve the recognition ability of the license plate recognition system is also a practical problem to be solved.

Photosensitive parts handling the external environment

The environment is the main factor affecting the recognition of the license plate. When the vehicle image is collected, the ambient light changes drastically, the daytime light is strong, the nighttime is weak, the surface light is different from the backlight, and the morning and afternoon light directions are different. The influence of ambient light is large, the vehicle speed is too high, and the dynamic range of the collection equipment makes the image quality difficult to be effectively guaranteed. When the recognition algorithm considers that the license plate has reached the optimal imaging position, the system triggers the system to start shooting, which has high requirements for the reliability and response speed of the trigger device. Therefore, to solve the problem of low recognition rate caused by the environment, it is also necessary to deal with the external environment by the photosensitive member of the camera.

Image preprocessing

Before the license plate is positioned, the image is generally pre-processed, and then the license plate is positioned, divided, and identified. The obtained license plate image may contain more noise, or the image contrast is not strong, the license plate is partially blocked, the license plate is stained, dirty, blurred, faded, other character area interference, and image blur caused by motion. Therefore, the positioning algorithm is more difficult to implement. For character segmentation, there may be unfavorable factors such as uneven illumination, severe stains, license plate tilt, low contrast, license plate fading, license plate character sticking, etc., so it is necessary to develop an algorithm suitable for it. If the algorithm can adapt to various complex environments and noise, license plate occlusion, license plate tilt, etc., then it can greatly improve the probability of license plate recognition.

As more and more people increase their purchasing power, many people drive every day, whether it is to go to work or a visitor. This also leads to the demand for parking lots becoming extra, if this time a parking lot If the license plate recognition system is faulty, it is not only the loss of the source of the passengers, so it is very important for some people who manage the parking lot to understand some basic causes of errors. So, why are some common identification errors in the license plate recognition system and how to solve them?

The difficulty of positioning the license plate area is mainly from the collected images. Due to the diversity of the collected license plate images, and the image is affected by many factors, such as rain, fog, light, etc., some license plate image quality appears different degrees. The difference.

Solution suggestion: This situation is generally only in the case of clear focus of the image, the recognition results can achieve a satisfactory result. That has certain requirements for the photosensitive parts of the parking system. The video camera will also need to improve some configurations. Look at the video settings are not normal (set normal), display resolution, color desktop size adjustment.

Whether using a ground coil or an infrared trigger, the impact of the model is obvious. The trigger position of the cart and the cart may exceed 0.5 meters even at low speeds. When it exceeds 1 meter, the captured image may not be the clearest. of. Solution suggestion: For this kind of situation, it is generally necessary to pre-process the image before the license plate is positioned, and then carry out the positioning, segmentation and recognition of the license plate, which is also very challenging for the system.

In the actual application process on the road and in the city, it is difficult to ensure that the license plates involved are not defaced, and the license plate will inevitably become polluted and worn after several years of use.

Solution suggestion: For this situation, many intelligent parking systems have gradually solved such problems, but if it is too serious, it has to rely on labor.

1), management software crashes

(1) Restart the computer

(2) The video card conflicts with another card, changing the slot

(3) Change the interrupt number

2), the software can not enter the access management

(1) Check if the card reader is turned on and the communication is normal.

(2) The network is normal

(3) The computer graphics card is set properly.

(4) Is the system setting normal?

3), the software responds slowly when the card is swiped or the computer freezes

(1) Check if the operating system is normal (reinstall the system)

(2) The vehicle database data cannot be too much (filed a small amount of data)

(3) Whether the network is stable and the transmission rate is low (adjustment speed)

(4) Is the workstation setting normal (unified setting)

4), the software image does not take pictures or can not retain the image

(1) Check if the mapped network drive letter is normal (reset)

(2) Is the image deleted?

(3) Check if the image is saved for 0 days

5), the computer is unstable, the system is often wrong, and it is easy to crash when operating the software.

(1) computer killing

(2) Check if the vehicle record recorded by the computer is too much

6), the computer network is unstable, when the time is disconnected

(1) RJ-4S head is bad and can be replaced

(2) The terminal resistance of the BNC head is broken, (replace)

(3) Is the power supply of the HUB stable?

(4) Whether the network card settings are changed

Of course, there are some problems with the recognition of the license plate recognition system. There are some objective reasons for the need for manual or enhanced equipment. This depends on the development of the parking system business in the future.

Box Header Connector

Box Header Connector,Box Header Add Housing Connectors,180 Degree Box Header Connector,1.27Mm Box Header Connector

Shenzhen Hongyian Electronics Co., Ltd. , https://www.hongyiancon.com