Research Of Intelligent Ground Lock Based On License Plate Recognition

The increase in car ownership is an inevitable phenomenon in the improvement of the country's comprehensive economic level.In recent years,the contradiction between the rapid growth of car ownership and the slow growth of parking spaces brings great pressure to the management of urban parking spaces.This paper proposes an intelligent ground lock based on license plate recognition.The ground lock innovatively introduces the license plate recognition function,which can accurately identify vehicle information and greatly improve the level of unmanned management of parking spaces.This intelligent ground lock has high investment value and can be widely used in many aspects: unmanned management of parking spaces in new energy charging stations,management of public parking area in the mall,management of a small number of parking space rental by private owners,etc.In this paper,the structure design and hardware selection of intelligent ground lock based on license plate recognition are carried out,and the software is designed with the knowledge of image technology and deep learning.The license plate recognition function is an innovation point and a difficult development point of intelligent ground lock.The development of license plate recognition function is divided into four parts: image preprocessing,license plate location and correction,character segmentation and character recognition.In the image preprocessing part,according to the actual needs of the product operation scene,the paper introduces the bilateral fuzzy algorithm to improve the sharpness of the image while noise reduction,and improve the efficiency of license plate recognition from the source;In the license plate location and correction part,the article uses an edge detection-based algorithm to accurately locate the license plate,and can use the texture field-based method to achieve character correction.The essence of the algorithm is to count the most densely textured directions in the license plate image,thereby determine the direction of license plate correction;In the character segmentation section,this paper uses the method of character sliding template to segment.Firstly,HSV color model is used to judge the background color of license plate,and then the specific sliding scheme is selected according to the background color of license plate,that is,if the 7-character sliding template is selected for blue,the 8-character sliding template is selected for green;In the part of character recognition,the paper uses a CNN Convolutional Neural Network model,using deep learning algorithms,and finally recognizes the segmented characters one by one through training and feedback layer by layer.After completing the design of software and hardware of intelligent ground lock,this paper tests 1000 groups of products in the actual application scenario of charging station,and the recognition success rate is up to 99.83% in the case of normal light.The actual use effect is good,which greatly optimizes the operation process of parking charging,and has strong application value.