Research On Key Techniques Of Vehicle License Plate Recognition Under Nature Scenes

Nowadays license plate recognition(LPR) became a key technique to intelligent transportation systems(ITS) such as parking lots access control, road traffic monitoring, automated payment of tolls on high ways, and security access. With the growing popularity of ITS, researching on more stable, rapid and accurate LPR technology has tremendous significance and social value.There are three key techniques of license plate recognition, including the license plate location, the vehicle character segmentation and character recognition. The article explored and studied them in depth, and raised some effective methods, specifically including the following:A new algorithm of vehicle license plate location based on edge detection- morphology-line scan is proposed. Firstly, the vertical edges was obtained by edge detection. Secondly, morphology operation was used to get the candidate plate regions. Finally, double line scan based on the license plate characteristics was used for plate region detection. The experimental results show that this method has higher veracity and stronger practicability.This paper presents a new algorithm for character segmentation. Firstly, characters top-bottom edges can be obtained by searching from middle to end. Secondly, generate the template of character string according to the prior knowledge and character height. Thirdly, get the left-right edges by optimization matching location using sliding the template and matching the vertical projection vector from left to right. And then, the single characters can be segmented using the left edge position and template characteristics. The experiment results show the good performance of the segmentation algorithm.Based on the analysis of the traditional template matching character recognition method's shortcomings, the paper proposed an improvement algorithm. New method make full use of the character's background for matching, effectively solved the fracture and fuzzy character recognition. And gives a similar character recognition solution. According to the complex Chinese characters, proposed a two-level-network recognition method. Experimental performance of the two methods proved very good.