First, analyze the principle of license plate recognition. License plate recognition is based on image segmentation and image recognition theory. It analyzes and processes the image containing the license plate recognition vehicle number plate, so as to determine the position of the license plate in the image, and further extract and recognize the text. character. The license plate recognition process includes a series of algorithm operations such as image acquisition, preprocessing, license plate positioning, character segmentation, character recognition, and result output. The operation process is shown in the following figure:
So how is the 99.7% license plate recognition rate achieved?
First of all, camera imaging is the basis of license plate recognition, and stable and high-quality imaging results lay a solid foundation for high recognition rate;
Secondly, on-site installation and debugging is an important guarantee for license plate recognition. The professional consulting service department provides professional and reasonable installation advice for each installation site. Angle, distance, fill light adjustment, etc. provide a strong guarantee for high license plate recognition rate. ;
Again, powerful recognition algorithms.
1. The core algorithm supports a variety of functions: supports a wide variety of license plates and flexible working modes.
The core algorithm of license plate recognition supports all kinds of license plates: ordinary blue card, single-layer yellow card, double-layer yellow card, police car license plate, armed police license plate, military license plate (new army license plate), embassy license plate, 02 personalized license plate, coach license plate, agricultural license plate , Trailer license plate, civil aviation license plate, Hong Kong and Macau entry and exit license plate, Taiwan license plate, etc.
Support coil trigger recognition work mode and video stream recognition work mode, flexible and complementary.
It supports body color recognition, car logo recognition, and model recognition.
2. Strong robustness
(Robustness means that it can work normally in abnormal scenes and abnormal conditions, and has a high recognition rate) For reflective license plates, backlight license plates, yin and yang license plates, deformed license plates, defaced license plates, Mercedes-Benz special license plates, inclined license plates, low Contrast license plates, license plates in rainy and foggy weather, over-exploded license plates, license plates with sticky borders, similar characters and other special cases have a high license plate recognition rate.
3. Algorithm real-time
In coil triggering mode, a single frame takes about 500ms; in video stream recognition mode, a single frame takes about 100ms.