Fuzzy license plate recognition method

Abstract

The invention discloses a fuzzy license plate recognition method, and the method comprises the following steps: S1, carrying out the inclined fuzzy license plate processing, and carrying out visual correction on an obtained license plate image in a perspective transformation mode; s2, segmenting license plate characters, concretely performing blind segmentation on the license plate according to the character positions in the license plate and the proportion of each character, and dividing the license plate into a plurality of parts; and S3, convolutional neural network identification by establishing a convolutional neural network to perform license plate character identification, and controlling the scale of the whole network through local connection, weight sharing and down-sampling on the structure so as to realize robustness of the network to the identified image in the aspect of deformation. According to the method, the processing time is short, the recognition rate of fuzzy license plates which cannot be recognized by human eyes or are difficult to recognize can be remarkably increased, the difficulties in the traffic law enforcement process are solved, and a favorable law enforcement basis is provided for law enforcement personnel.

Description

technical field [0001] The invention relates to a recognition method, in particular to a deep learning-based fuzzy license plate recognition method, which belongs to the technical field of artificial intelligence. Background technique [0002] With the continuous development of the automobile industry, the number of vehicles in the world continues to increase. As the license plate number used to identify the unique "identity" of each vehicle, its practicality and importance are also increasing day by day. [0003] In such an era, license plate recognition technology emerged as the times require. As an important technology in the fields of transportation and security, license plate recognition can realize automatic registration and verification of vehicle "identity" without any changes to the vehicle. This technology has been widely used in highway toll collection, parking management, weighing system, traffic guidance, traffic law enforcement, highway inspection, vehicle sch...

AI Technical Summary

Problems solved by technology

[0005] However, through long-term practice, technicians have found that the above-mentioned methods are extremely susceptible to interference, and it takes a long time to recognize a single character on a license plate, and there are certain errors in the recognition accuracy.

In addition, in the actual application environment, due to the rapid movement of the vehicle, it is easy to cause problems such as blurring and ghosting in the collected vehicle images, which further increases the difficulty of license plate recognition and makes it impossible to obtain accurate license plate information.

Images