Recognition Optimization of License Plate Targets Based on Improved Neural Network Model

With the continuous development of social economy, private cars have become more and more, and the traffic pressure is also increasing. To more accurately recognize vehicles violating traffic rules, the problem of recognition and optimization of license plate targets has become an urgent task with certain practical significance and guiding significance. In this paper, the license plate recognition (LPR) system is taken as the main research object, which is improved and optimized. Firstly, an improved neural network model is established by applying the improved convolutional neural network (CNN) algorithm to the license plate recognition system and based on the improved activation function, and the relevant experimental results are obtained; meanwhile compared with the traditional CNN model, the corresponding experimental results are obtained; finally, the experimental conclusions are obtained, i.e., the LPR system using improved CNN algorithm has better performance in recognizing license plate target and can more accurately recognize licence plates which are shielded. Therefore, the improved neural network model has great development potential in the application of the LPR system.