License Plate Recognition Technology Analysis

The license plate recognition is a pattern recognition technology that uses the dynamic video or static image of the vehicle to automatically identify the license plate number and license plate color. The core technologies include license plate location algorithm, license plate character segmentation algorithm and optical character recognition algorithm.

License plate recognition technology works

Vehicle detection: It can detect the passage of the vehicle through various methods such as buried coil detection, infrared detection, radar detection technology, and video detection, and triggers image capture and capture.

Image acquisition: Real-time, continuous recording and acquisition of passing vehicles through high-definition video capture.

Preprocessing: noise filtering, auto white balance, auto exposure and gamma correction, edge enhancement, contrast adjustment, etc.

License Plate Positioning: Rows are scanned on the grayscale image after image preprocessing to determine the license plate area.

Character segmentation: After the license plate area is located in the image, the character area is accurately positioned through processing such as graying and binarization, and then character segmentation is performed according to the character size feature.

Character recognition: The segmented characters are scaled, extracted, and matched with the standard character representation in the character database template.

Result output: The license plate recognition result is output in text format.

License Plate Recognition Technology Workflow

The license plate recognition system adopts a highly modular design, which regards each link of the license plate recognition process as an independent module.

First, vehicle detection and tracking module

The vehicle detection and tracking module mainly analyzes the video stream to determine the location of the vehicle, tracks the vehicle in the image, and records the close-up picture of the vehicle at the best moment of the vehicle location. The system is very good due to the addition of the tracking module. To overcome various external disturbances, to make more reasonable identification results, it is possible to detect unlicensed vehicles and output the results.

Second, the license plate positioning module

The license plate positioning module is a very important link and is the basis for follow-up links. Its accuracy has a great impact on the overall system performance. The license plate system completely abandons the previous algorithm idea and realizes a new license plate positioning algorithm based on multiple feature fusion based on learning. It is suitable for various complicated background environments and different camera angles.

Third, license plate correction and precision positioning module

Due to the limitation of shooting conditions, the license plate in the image always has a certain tilt, and it needs a correction and fine positioning to further improve the quality of the license plate image, and prepares the segmentation and recognition module. The use of well-designed fast image processing filters not only makes calculations faster, but also uses the overall information of the license plate, avoiding the effects of local noise. Another advantage of using this algorithm is that the analysis of multiple intermediate results can also fine-tune the license plate, further reducing the impact of non-license area.

Fourth, license plate segmentation module

The license plate segmentation module of the license plate system utilizes various features such as grayscale, color, and edge distribution of the license plate text, which can better suppress the influence of other noise around the license plate and can tolerate the license plate with a certain inclination angle. This algorithm is advantageous for applications such as mobile auditing, where the license plate image is noisier.

Fifth, license plate recognition module

In the license plate recognition system, multiple recognition models are commonly used to identify license plates and a hierarchical character recognition process is constructed, which can effectively improve the accuracy of character recognition. On the other hand, prior to character recognition, the use of computer intelligence algorithms to perform pre-processing on character images can not only preserve image information as much as possible, but also improve image quality, improve the distinguishability of similar characters, and ensure the reliability of character recognition.

Sixth, license plate recognition result decision module

The recognition result decision module, specifically, the decision module uses an historical record left by the license plate through the field of vision to make an intelligent decision on the recognition result. It calculates the comprehensive credibility of the license plate by calculating the number of observed frames, the stability of the recognition result, the stability of the track, the stability of the speed, the average reliability, and the similarity, and determines whether to continue tracking the license plate or output. Identify the result or reject the result. This method uses all frame information comprehensively, reduces the accidental errors brought about by previous recognition algorithms based on a single image, and greatly improves the system's recognition rate and the correctness and reliability of the recognition results.

Seventh, license plate tracking module

The license plate tracking module records various historical information such as the position and appearance, recognition result, and credibility of the vehicle license plate in each frame during the running of the vehicle. Because the license plate tracking module adopts a movement model and updating model with certain fault tolerance, license plates that are covered for a short time or are instantly blurred can still be tracked and predicted correctly, and only one recognition result is output finally.