Research On License Plate Location And Image Restoration Technology Of Defective License Plates In Non-Specific Scenes

As the key technology in intelligent transportation systems,license plate recognition has always been a research hotspot of researchers.The current license plate detection technology has been widely used in our daily life.In the case of non-specific scenes or stained license plates,the recognition effect will be greatly reduced.It is found that there are two main reasons for the poor recognition effect of license plates in non-specific scenes:poor license plate performance and poor image quality.Starting from the above problems,this paper introduces the improvement and innovation to improve the license plate detection effect in non-specific scenes from the perspectives of license plate localization and itnage repair in the process of license plate detection.The main work of this article is as follows:(1)In order to reduce the problem of low recognition accuracy caused by image quality,this paper studied and analyzed the traditional image repair algorithm and generated antagonistic network.According to the characteristics of license plate image,a license plate image repair and enhancement method based on generated antagonistic network was proposed.This method starts from the image edge information,and while preserving the license plate image feature information,it adopts a multi-stage method to improve the image quality of the license plate image.(2)In order to reduce the problem of low recognition accuracy caused by image quality,this paper studies and analyzes traditional image repair algorithms and generative confrontation networks,and proposes a method for image license image enhancement based on generative confrontation networks.The edge and texture aspects respectively restore the missing information of the image.In addition,by adding the structural loss to the loss function of the model,the image generated by the generator is more visually authentic.The experiments proved that the algorithm can effectively improve the recognition accuracy of the damaged license plates.The above work has studied the license plate positioning and the repair and enhancement of the license plate image,and completed the repair of the stained and blurred license plate image,which has reference value and certain practical value for the research of license plate recognition in harsh environments.