A License Plate Recognition System Based On Deep Learning Method

In recent years,with the development and progress of society,all kinds of cars have entered the homes of ordinary people,and people also realize the importance of car license plate.In view of the shortcomings of traditional license plate recognition methods,such as instability and complex process,prone to errors,and the advantages of license plate recognition method based on deep learning,such as strong robustness and simple operation,this paper proposes a license plate location algorithm based on color features and edge detection,and designs a license plate recognition system based on improved alexnet network.In the first part,after preprocessing the license plate image,several license plate location methods which are widely used at present are analyzed and compared.Finally,a new license plate location method based on the combination of color segmentation and edge detection is proposed,and support vector machine is added In order to improve the accuracy,SVM binary classification model is used to further screen the license plate regions.Finally,the tilt correction algorithm based on Open CV affine transform is used to correct the license plate.The experiment shows that this method can complete the task of license plate location and detection.In the part of license plate recognition,based on the investigation of various classic convolutional neural networks,this paper makes a comparative experiment of the main network.According to the specific technical requirements of license plate recognition,especially considering the real-time requirements of time,this paper chooses to improve the classic alexnet convolutional neural network,and introduces the principle of multi label classification,and uses softmax classifier for recognition,A license plate recognition system based on deep learning method is proposed.Finally,the comparison of the recognition accuracy of the traditional and the proposed license plate recognition system in complex environment and the comparison of typical cases are added.The results show that the license plate recognition based on deep learning can improve the recognition accuracy.